swapfile.c 77.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
/*
 *  linux/mm/swapfile.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *  Swap reorganised 29.12.95, Stephen Tweedie
 */

#include <linux/mm.h>
#include <linux/hugetlb.h>
#include <linux/mman.h>
#include <linux/slab.h>
#include <linux/kernel_stat.h>
#include <linux/swap.h>
#include <linux/vmalloc.h>
#include <linux/pagemap.h>
#include <linux/namei.h>
17
#include <linux/shmem_fs.h>
L
Linus Torvalds 已提交
18
#include <linux/blkdev.h>
19
#include <linux/random.h>
L
Linus Torvalds 已提交
20 21 22 23
#include <linux/writeback.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/init.h>
H
Hugh Dickins 已提交
24
#include <linux/ksm.h>
L
Linus Torvalds 已提交
25 26 27
#include <linux/rmap.h>
#include <linux/security.h>
#include <linux/backing-dev.h>
I
Ingo Molnar 已提交
28
#include <linux/mutex.h>
29
#include <linux/capability.h>
L
Linus Torvalds 已提交
30
#include <linux/syscalls.h>
31
#include <linux/memcontrol.h>
K
Kay Sievers 已提交
32
#include <linux/poll.h>
33
#include <linux/oom.h>
34 35
#include <linux/frontswap.h>
#include <linux/swapfile.h>
36
#include <linux/export.h>
L
Linus Torvalds 已提交
37 38 39 40

#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <linux/swapops.h>
41
#include <linux/swap_cgroup.h>
L
Linus Torvalds 已提交
42

H
Hugh Dickins 已提交
43 44 45
static bool swap_count_continued(struct swap_info_struct *, pgoff_t,
				 unsigned char);
static void free_swap_count_continuations(struct swap_info_struct *);
46
static sector_t map_swap_entry(swp_entry_t, struct block_device**);
H
Hugh Dickins 已提交
47

48
DEFINE_SPINLOCK(swap_lock);
A
Adrian Bunk 已提交
49
static unsigned int nr_swapfiles;
50 51
atomic_long_t nr_swap_pages;
/* protected with swap_lock. reading in vm_swap_full() doesn't need lock */
L
Linus Torvalds 已提交
52
long total_swap_pages;
53
static int least_priority;
L
Linus Torvalds 已提交
54 55 56 57 58 59

static const char Bad_file[] = "Bad swap file entry ";
static const char Unused_file[] = "Unused swap file entry ";
static const char Bad_offset[] = "Bad swap offset entry ";
static const char Unused_offset[] = "Unused swap offset entry ";

60 61 62 63
/*
 * all active swap_info_structs
 * protected with swap_lock, and ordered by priority.
 */
64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79
PLIST_HEAD(swap_active_head);

/*
 * all available (active, not full) swap_info_structs
 * protected with swap_avail_lock, ordered by priority.
 * This is used by get_swap_page() instead of swap_active_head
 * because swap_active_head includes all swap_info_structs,
 * but get_swap_page() doesn't need to look at full ones.
 * This uses its own lock instead of swap_lock because when a
 * swap_info_struct changes between not-full/full, it needs to
 * add/remove itself to/from this list, but the swap_info_struct->lock
 * is held and the locking order requires swap_lock to be taken
 * before any swap_info_struct->lock.
 */
static PLIST_HEAD(swap_avail_head);
static DEFINE_SPINLOCK(swap_avail_lock);
L
Linus Torvalds 已提交
80

81
struct swap_info_struct *swap_info[MAX_SWAPFILES];
L
Linus Torvalds 已提交
82

I
Ingo Molnar 已提交
83
static DEFINE_MUTEX(swapon_mutex);
L
Linus Torvalds 已提交
84

K
Kay Sievers 已提交
85 86 87 88
static DECLARE_WAIT_QUEUE_HEAD(proc_poll_wait);
/* Activity counter to indicate that a swapon or swapoff has occurred */
static atomic_t proc_poll_event = ATOMIC_INIT(0);

89
static inline unsigned char swap_count(unsigned char ent)
90
{
H
Hugh Dickins 已提交
91
	return ent & ~SWAP_HAS_CACHE;	/* may include SWAP_HAS_CONT flag */
92 93
}

94
/* returns 1 if swap entry is freed */
95 96 97
static int
__try_to_reclaim_swap(struct swap_info_struct *si, unsigned long offset)
{
98
	swp_entry_t entry = swp_entry(si->type, offset);
99 100 101
	struct page *page;
	int ret = 0;

102
	page = find_get_page(swap_address_space(entry), entry.val);
103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118
	if (!page)
		return 0;
	/*
	 * This function is called from scan_swap_map() and it's called
	 * by vmscan.c at reclaiming pages. So, we hold a lock on a page, here.
	 * We have to use trylock for avoiding deadlock. This is a special
	 * case and you should use try_to_free_swap() with explicit lock_page()
	 * in usual operations.
	 */
	if (trylock_page(page)) {
		ret = try_to_free_swap(page);
		unlock_page(page);
	}
	page_cache_release(page);
	return ret;
}
119

120 121 122 123 124 125 126
/*
 * swapon tell device that all the old swap contents can be discarded,
 * to allow the swap device to optimize its wear-levelling.
 */
static int discard_swap(struct swap_info_struct *si)
{
	struct swap_extent *se;
127 128
	sector_t start_block;
	sector_t nr_blocks;
129 130
	int err = 0;

131 132 133 134 135 136
	/* Do not discard the swap header page! */
	se = &si->first_swap_extent;
	start_block = (se->start_block + 1) << (PAGE_SHIFT - 9);
	nr_blocks = ((sector_t)se->nr_pages - 1) << (PAGE_SHIFT - 9);
	if (nr_blocks) {
		err = blkdev_issue_discard(si->bdev, start_block,
137
				nr_blocks, GFP_KERNEL, 0);
138 139 140 141
		if (err)
			return err;
		cond_resched();
	}
142

143 144 145
	list_for_each_entry(se, &si->first_swap_extent.list, list) {
		start_block = se->start_block << (PAGE_SHIFT - 9);
		nr_blocks = (sector_t)se->nr_pages << (PAGE_SHIFT - 9);
146 147

		err = blkdev_issue_discard(si->bdev, start_block,
148
				nr_blocks, GFP_KERNEL, 0);
149 150 151 152 153 154 155 156
		if (err)
			break;

		cond_resched();
	}
	return err;		/* That will often be -EOPNOTSUPP */
}

157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
/*
 * swap allocation tell device that a cluster of swap can now be discarded,
 * to allow the swap device to optimize its wear-levelling.
 */
static void discard_swap_cluster(struct swap_info_struct *si,
				 pgoff_t start_page, pgoff_t nr_pages)
{
	struct swap_extent *se = si->curr_swap_extent;
	int found_extent = 0;

	while (nr_pages) {
		if (se->start_page <= start_page &&
		    start_page < se->start_page + se->nr_pages) {
			pgoff_t offset = start_page - se->start_page;
			sector_t start_block = se->start_block + offset;
172
			sector_t nr_blocks = se->nr_pages - offset;
173 174 175 176 177 178 179 180 181 182 183 184

			if (nr_blocks > nr_pages)
				nr_blocks = nr_pages;
			start_page += nr_blocks;
			nr_pages -= nr_blocks;

			if (!found_extent++)
				si->curr_swap_extent = se;

			start_block <<= PAGE_SHIFT - 9;
			nr_blocks <<= PAGE_SHIFT - 9;
			if (blkdev_issue_discard(si->bdev, start_block,
185
				    nr_blocks, GFP_NOIO, 0))
186 187 188
				break;
		}

189
		se = list_next_entry(se, list);
190 191 192
	}
}

193 194 195
#define SWAPFILE_CLUSTER	256
#define LATENCY_LIMIT		256

196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253
static inline void cluster_set_flag(struct swap_cluster_info *info,
	unsigned int flag)
{
	info->flags = flag;
}

static inline unsigned int cluster_count(struct swap_cluster_info *info)
{
	return info->data;
}

static inline void cluster_set_count(struct swap_cluster_info *info,
				     unsigned int c)
{
	info->data = c;
}

static inline void cluster_set_count_flag(struct swap_cluster_info *info,
					 unsigned int c, unsigned int f)
{
	info->flags = f;
	info->data = c;
}

static inline unsigned int cluster_next(struct swap_cluster_info *info)
{
	return info->data;
}

static inline void cluster_set_next(struct swap_cluster_info *info,
				    unsigned int n)
{
	info->data = n;
}

static inline void cluster_set_next_flag(struct swap_cluster_info *info,
					 unsigned int n, unsigned int f)
{
	info->flags = f;
	info->data = n;
}

static inline bool cluster_is_free(struct swap_cluster_info *info)
{
	return info->flags & CLUSTER_FLAG_FREE;
}

static inline bool cluster_is_null(struct swap_cluster_info *info)
{
	return info->flags & CLUSTER_FLAG_NEXT_NULL;
}

static inline void cluster_set_null(struct swap_cluster_info *info)
{
	info->flags = CLUSTER_FLAG_NEXT_NULL;
	info->data = 0;
}

S
Shaohua Li 已提交
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337
/* Add a cluster to discard list and schedule it to do discard */
static void swap_cluster_schedule_discard(struct swap_info_struct *si,
		unsigned int idx)
{
	/*
	 * If scan_swap_map() can't find a free cluster, it will check
	 * si->swap_map directly. To make sure the discarding cluster isn't
	 * taken by scan_swap_map(), mark the swap entries bad (occupied). It
	 * will be cleared after discard
	 */
	memset(si->swap_map + idx * SWAPFILE_CLUSTER,
			SWAP_MAP_BAD, SWAPFILE_CLUSTER);

	if (cluster_is_null(&si->discard_cluster_head)) {
		cluster_set_next_flag(&si->discard_cluster_head,
						idx, 0);
		cluster_set_next_flag(&si->discard_cluster_tail,
						idx, 0);
	} else {
		unsigned int tail = cluster_next(&si->discard_cluster_tail);
		cluster_set_next(&si->cluster_info[tail], idx);
		cluster_set_next_flag(&si->discard_cluster_tail,
						idx, 0);
	}

	schedule_work(&si->discard_work);
}

/*
 * Doing discard actually. After a cluster discard is finished, the cluster
 * will be added to free cluster list. caller should hold si->lock.
*/
static void swap_do_scheduled_discard(struct swap_info_struct *si)
{
	struct swap_cluster_info *info;
	unsigned int idx;

	info = si->cluster_info;

	while (!cluster_is_null(&si->discard_cluster_head)) {
		idx = cluster_next(&si->discard_cluster_head);

		cluster_set_next_flag(&si->discard_cluster_head,
						cluster_next(&info[idx]), 0);
		if (cluster_next(&si->discard_cluster_tail) == idx) {
			cluster_set_null(&si->discard_cluster_head);
			cluster_set_null(&si->discard_cluster_tail);
		}
		spin_unlock(&si->lock);

		discard_swap_cluster(si, idx * SWAPFILE_CLUSTER,
				SWAPFILE_CLUSTER);

		spin_lock(&si->lock);
		cluster_set_flag(&info[idx], CLUSTER_FLAG_FREE);
		if (cluster_is_null(&si->free_cluster_head)) {
			cluster_set_next_flag(&si->free_cluster_head,
						idx, 0);
			cluster_set_next_flag(&si->free_cluster_tail,
						idx, 0);
		} else {
			unsigned int tail;

			tail = cluster_next(&si->free_cluster_tail);
			cluster_set_next(&info[tail], idx);
			cluster_set_next_flag(&si->free_cluster_tail,
						idx, 0);
		}
		memset(si->swap_map + idx * SWAPFILE_CLUSTER,
				0, SWAPFILE_CLUSTER);
	}
}

static void swap_discard_work(struct work_struct *work)
{
	struct swap_info_struct *si;

	si = container_of(work, struct swap_info_struct, discard_work);

	spin_lock(&si->lock);
	swap_do_scheduled_discard(si);
	spin_unlock(&si->lock);
}

338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
/*
 * The cluster corresponding to page_nr will be used. The cluster will be
 * removed from free cluster list and its usage counter will be increased.
 */
static void inc_cluster_info_page(struct swap_info_struct *p,
	struct swap_cluster_info *cluster_info, unsigned long page_nr)
{
	unsigned long idx = page_nr / SWAPFILE_CLUSTER;

	if (!cluster_info)
		return;
	if (cluster_is_free(&cluster_info[idx])) {
		VM_BUG_ON(cluster_next(&p->free_cluster_head) != idx);
		cluster_set_next_flag(&p->free_cluster_head,
			cluster_next(&cluster_info[idx]), 0);
		if (cluster_next(&p->free_cluster_tail) == idx) {
			cluster_set_null(&p->free_cluster_tail);
			cluster_set_null(&p->free_cluster_head);
		}
		cluster_set_count_flag(&cluster_info[idx], 0, 0);
	}

	VM_BUG_ON(cluster_count(&cluster_info[idx]) >= SWAPFILE_CLUSTER);
	cluster_set_count(&cluster_info[idx],
		cluster_count(&cluster_info[idx]) + 1);
}

/*
 * The cluster corresponding to page_nr decreases one usage. If the usage
 * counter becomes 0, which means no page in the cluster is in using, we can
 * optionally discard the cluster and add it to free cluster list.
 */
static void dec_cluster_info_page(struct swap_info_struct *p,
	struct swap_cluster_info *cluster_info, unsigned long page_nr)
{
	unsigned long idx = page_nr / SWAPFILE_CLUSTER;

	if (!cluster_info)
		return;

	VM_BUG_ON(cluster_count(&cluster_info[idx]) == 0);
	cluster_set_count(&cluster_info[idx],
		cluster_count(&cluster_info[idx]) - 1);

	if (cluster_count(&cluster_info[idx]) == 0) {
S
Shaohua Li 已提交
383 384 385 386 387
		/*
		 * If the swap is discardable, prepare discard the cluster
		 * instead of free it immediately. The cluster will be freed
		 * after discard.
		 */
388 389
		if ((p->flags & (SWP_WRITEOK | SWP_PAGE_DISCARD)) ==
				 (SWP_WRITEOK | SWP_PAGE_DISCARD)) {
S
Shaohua Li 已提交
390 391 392 393
			swap_cluster_schedule_discard(p, idx);
			return;
		}

394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
		cluster_set_flag(&cluster_info[idx], CLUSTER_FLAG_FREE);
		if (cluster_is_null(&p->free_cluster_head)) {
			cluster_set_next_flag(&p->free_cluster_head, idx, 0);
			cluster_set_next_flag(&p->free_cluster_tail, idx, 0);
		} else {
			unsigned int tail = cluster_next(&p->free_cluster_tail);
			cluster_set_next(&cluster_info[tail], idx);
			cluster_set_next_flag(&p->free_cluster_tail, idx, 0);
		}
	}
}

/*
 * It's possible scan_swap_map() uses a free cluster in the middle of free
 * cluster list. Avoiding such abuse to avoid list corruption.
 */
410 411
static bool
scan_swap_map_ssd_cluster_conflict(struct swap_info_struct *si,
412 413
	unsigned long offset)
{
414 415 416
	struct percpu_cluster *percpu_cluster;
	bool conflict;

417
	offset /= SWAPFILE_CLUSTER;
418
	conflict = !cluster_is_null(&si->free_cluster_head) &&
419 420
		offset != cluster_next(&si->free_cluster_head) &&
		cluster_is_free(&si->cluster_info[offset]);
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481

	if (!conflict)
		return false;

	percpu_cluster = this_cpu_ptr(si->percpu_cluster);
	cluster_set_null(&percpu_cluster->index);
	return true;
}

/*
 * Try to get a swap entry from current cpu's swap entry pool (a cluster). This
 * might involve allocating a new cluster for current CPU too.
 */
static void scan_swap_map_try_ssd_cluster(struct swap_info_struct *si,
	unsigned long *offset, unsigned long *scan_base)
{
	struct percpu_cluster *cluster;
	bool found_free;
	unsigned long tmp;

new_cluster:
	cluster = this_cpu_ptr(si->percpu_cluster);
	if (cluster_is_null(&cluster->index)) {
		if (!cluster_is_null(&si->free_cluster_head)) {
			cluster->index = si->free_cluster_head;
			cluster->next = cluster_next(&cluster->index) *
					SWAPFILE_CLUSTER;
		} else if (!cluster_is_null(&si->discard_cluster_head)) {
			/*
			 * we don't have free cluster but have some clusters in
			 * discarding, do discard now and reclaim them
			 */
			swap_do_scheduled_discard(si);
			*scan_base = *offset = si->cluster_next;
			goto new_cluster;
		} else
			return;
	}

	found_free = false;

	/*
	 * Other CPUs can use our cluster if they can't find a free cluster,
	 * check if there is still free entry in the cluster
	 */
	tmp = cluster->next;
	while (tmp < si->max && tmp < (cluster_next(&cluster->index) + 1) *
	       SWAPFILE_CLUSTER) {
		if (!si->swap_map[tmp]) {
			found_free = true;
			break;
		}
		tmp++;
	}
	if (!found_free) {
		cluster_set_null(&cluster->index);
		goto new_cluster;
	}
	cluster->next = tmp + 1;
	*offset = tmp;
	*scan_base = tmp;
482 483
}

484 485
static unsigned long scan_swap_map(struct swap_info_struct *si,
				   unsigned char usage)
L
Linus Torvalds 已提交
486
{
487
	unsigned long offset;
488
	unsigned long scan_base;
489
	unsigned long last_in_cluster = 0;
490
	int latency_ration = LATENCY_LIMIT;
491

492
	/*
493 494 495 496 497 498 499
	 * We try to cluster swap pages by allocating them sequentially
	 * in swap.  Once we've allocated SWAPFILE_CLUSTER pages this
	 * way, however, we resort to first-free allocation, starting
	 * a new cluster.  This prevents us from scattering swap pages
	 * all over the entire swap partition, so that we reduce
	 * overall disk seek times between swap pages.  -- sct
	 * But we do now try to find an empty cluster.  -Andrea
500
	 * And we let swap pages go all over an SSD partition.  Hugh
501 502
	 */

503
	si->flags += SWP_SCANNING;
504
	scan_base = offset = si->cluster_next;
505

506 507 508 509 510 511
	/* SSD algorithm */
	if (si->cluster_info) {
		scan_swap_map_try_ssd_cluster(si, &offset, &scan_base);
		goto checks;
	}

512 513 514 515 516
	if (unlikely(!si->cluster_nr--)) {
		if (si->pages - si->inuse_pages < SWAPFILE_CLUSTER) {
			si->cluster_nr = SWAPFILE_CLUSTER - 1;
			goto checks;
		}
517

518
		spin_unlock(&si->lock);
519

520 521 522
		/*
		 * If seek is expensive, start searching for new cluster from
		 * start of partition, to minimize the span of allocated swap.
523 524
		 * If seek is cheap, that is the SWP_SOLIDSTATE si->cluster_info
		 * case, just handled by scan_swap_map_try_ssd_cluster() above.
525
		 */
526
		scan_base = offset = si->lowest_bit;
527 528 529 530
		last_in_cluster = offset + SWAPFILE_CLUSTER - 1;

		/* Locate the first empty (unaligned) cluster */
		for (; last_in_cluster <= si->highest_bit; offset++) {
L
Linus Torvalds 已提交
531
			if (si->swap_map[offset])
532 533
				last_in_cluster = offset + SWAPFILE_CLUSTER;
			else if (offset == last_in_cluster) {
534
				spin_lock(&si->lock);
535 536 537
				offset -= SWAPFILE_CLUSTER - 1;
				si->cluster_next = offset;
				si->cluster_nr = SWAPFILE_CLUSTER - 1;
538 539 540 541 542 543 544 545 546
				goto checks;
			}
			if (unlikely(--latency_ration < 0)) {
				cond_resched();
				latency_ration = LATENCY_LIMIT;
			}
		}

		offset = scan_base;
547
		spin_lock(&si->lock);
548
		si->cluster_nr = SWAPFILE_CLUSTER - 1;
L
Linus Torvalds 已提交
549
	}
550

551
checks:
552 553 554 555
	if (si->cluster_info) {
		while (scan_swap_map_ssd_cluster_conflict(si, offset))
			scan_swap_map_try_ssd_cluster(si, &offset, &scan_base);
	}
556
	if (!(si->flags & SWP_WRITEOK))
557
		goto no_page;
558 559
	if (!si->highest_bit)
		goto no_page;
560
	if (offset > si->highest_bit)
561
		scan_base = offset = si->lowest_bit;
562

563 564
	/* reuse swap entry of cache-only swap if not busy. */
	if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
565
		int swap_was_freed;
566
		spin_unlock(&si->lock);
567
		swap_was_freed = __try_to_reclaim_swap(si, offset);
568
		spin_lock(&si->lock);
569 570 571 572 573 574
		/* entry was freed successfully, try to use this again */
		if (swap_was_freed)
			goto checks;
		goto scan; /* check next one */
	}

575 576 577 578 579 580 581 582 583 584 585
	if (si->swap_map[offset])
		goto scan;

	if (offset == si->lowest_bit)
		si->lowest_bit++;
	if (offset == si->highest_bit)
		si->highest_bit--;
	si->inuse_pages++;
	if (si->inuse_pages == si->pages) {
		si->lowest_bit = si->max;
		si->highest_bit = 0;
586 587 588
		spin_lock(&swap_avail_lock);
		plist_del(&si->avail_list, &swap_avail_head);
		spin_unlock(&swap_avail_lock);
L
Linus Torvalds 已提交
589
	}
H
Hugh Dickins 已提交
590
	si->swap_map[offset] = usage;
591
	inc_cluster_info_page(si, si->cluster_info, offset);
592 593
	si->cluster_next = offset + 1;
	si->flags -= SWP_SCANNING;
594

595
	return offset;
596

597
scan:
598
	spin_unlock(&si->lock);
599
	while (++offset <= si->highest_bit) {
600
		if (!si->swap_map[offset]) {
601
			spin_lock(&si->lock);
602 603
			goto checks;
		}
604
		if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
605
			spin_lock(&si->lock);
606 607
			goto checks;
		}
608 609 610 611
		if (unlikely(--latency_ration < 0)) {
			cond_resched();
			latency_ration = LATENCY_LIMIT;
		}
612
	}
613
	offset = si->lowest_bit;
614
	while (offset < scan_base) {
615
		if (!si->swap_map[offset]) {
616
			spin_lock(&si->lock);
617 618
			goto checks;
		}
619
		if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
620
			spin_lock(&si->lock);
621 622
			goto checks;
		}
623 624 625 626
		if (unlikely(--latency_ration < 0)) {
			cond_resched();
			latency_ration = LATENCY_LIMIT;
		}
627
		offset++;
628
	}
629
	spin_lock(&si->lock);
630 631

no_page:
632
	si->flags -= SWP_SCANNING;
L
Linus Torvalds 已提交
633 634 635 636 637
	return 0;
}

swp_entry_t get_swap_page(void)
{
638
	struct swap_info_struct *si, *next;
639
	pgoff_t offset;
L
Linus Torvalds 已提交
640

641
	if (atomic_long_read(&nr_swap_pages) <= 0)
642
		goto noswap;
643
	atomic_long_dec(&nr_swap_pages);
644

645 646 647 648 649 650 651
	spin_lock(&swap_avail_lock);

start_over:
	plist_for_each_entry_safe(si, next, &swap_avail_head, avail_list) {
		/* requeue si to after same-priority siblings */
		plist_requeue(&si->avail_list, &swap_avail_head);
		spin_unlock(&swap_avail_lock);
652
		spin_lock(&si->lock);
653
		if (!si->highest_bit || !(si->flags & SWP_WRITEOK)) {
654 655 656 657 658 659 660 661 662 663 664 665
			spin_lock(&swap_avail_lock);
			if (plist_node_empty(&si->avail_list)) {
				spin_unlock(&si->lock);
				goto nextsi;
			}
			WARN(!si->highest_bit,
			     "swap_info %d in list but !highest_bit\n",
			     si->type);
			WARN(!(si->flags & SWP_WRITEOK),
			     "swap_info %d in list but !SWP_WRITEOK\n",
			     si->type);
			plist_del(&si->avail_list, &swap_avail_head);
666
			spin_unlock(&si->lock);
667
			goto nextsi;
668
		}
669

670
		/* This is called for allocating swap entry for cache */
H
Hugh Dickins 已提交
671
		offset = scan_swap_map(si, SWAP_HAS_CACHE);
672 673
		spin_unlock(&si->lock);
		if (offset)
674
			return swp_entry(si->type, offset);
675 676 677 678
		pr_debug("scan_swap_map of si %d failed to find offset\n",
		       si->type);
		spin_lock(&swap_avail_lock);
nextsi:
679 680 681 682
		/*
		 * if we got here, it's likely that si was almost full before,
		 * and since scan_swap_map() can drop the si->lock, multiple
		 * callers probably all tried to get a page from the same si
683 684 685 686 687
		 * and it filled up before we could get one; or, the si filled
		 * up between us dropping swap_avail_lock and taking si->lock.
		 * Since we dropped the swap_avail_lock, the swap_avail_head
		 * list may have been modified; so if next is still in the
		 * swap_avail_head list then try it, otherwise start over.
688
		 */
689 690
		if (plist_node_empty(&next->avail_list))
			goto start_over;
L
Linus Torvalds 已提交
691
	}
692

693 694
	spin_unlock(&swap_avail_lock);

695
	atomic_long_inc(&nr_swap_pages);
696 697
noswap:
	return (swp_entry_t) {0};
L
Linus Torvalds 已提交
698 699
}

S
Seth Jennings 已提交
700
/* The only caller of this function is now suspend routine */
701 702 703 704 705 706
swp_entry_t get_swap_page_of_type(int type)
{
	struct swap_info_struct *si;
	pgoff_t offset;

	si = swap_info[type];
707
	spin_lock(&si->lock);
708
	if (si && (si->flags & SWP_WRITEOK)) {
709
		atomic_long_dec(&nr_swap_pages);
710 711 712
		/* This is called for allocating swap entry, not cache */
		offset = scan_swap_map(si, 1);
		if (offset) {
713
			spin_unlock(&si->lock);
714 715
			return swp_entry(type, offset);
		}
716
		atomic_long_inc(&nr_swap_pages);
717
	}
718
	spin_unlock(&si->lock);
719 720 721
	return (swp_entry_t) {0};
}

722
static struct swap_info_struct *swap_info_get(swp_entry_t entry)
L
Linus Torvalds 已提交
723
{
724
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
725 726 727 728 729 730 731
	unsigned long offset, type;

	if (!entry.val)
		goto out;
	type = swp_type(entry);
	if (type >= nr_swapfiles)
		goto bad_nofile;
732
	p = swap_info[type];
L
Linus Torvalds 已提交
733 734 735 736 737 738 739
	if (!(p->flags & SWP_USED))
		goto bad_device;
	offset = swp_offset(entry);
	if (offset >= p->max)
		goto bad_offset;
	if (!p->swap_map[offset])
		goto bad_free;
740
	spin_lock(&p->lock);
L
Linus Torvalds 已提交
741 742 743
	return p;

bad_free:
744
	pr_err("swap_free: %s%08lx\n", Unused_offset, entry.val);
L
Linus Torvalds 已提交
745 746
	goto out;
bad_offset:
747
	pr_err("swap_free: %s%08lx\n", Bad_offset, entry.val);
L
Linus Torvalds 已提交
748 749
	goto out;
bad_device:
750
	pr_err("swap_free: %s%08lx\n", Unused_file, entry.val);
L
Linus Torvalds 已提交
751 752
	goto out;
bad_nofile:
753
	pr_err("swap_free: %s%08lx\n", Bad_file, entry.val);
L
Linus Torvalds 已提交
754 755
out:
	return NULL;
756
}
L
Linus Torvalds 已提交
757

758 759
static unsigned char swap_entry_free(struct swap_info_struct *p,
				     swp_entry_t entry, unsigned char usage)
L
Linus Torvalds 已提交
760
{
H
Hugh Dickins 已提交
761
	unsigned long offset = swp_offset(entry);
762 763
	unsigned char count;
	unsigned char has_cache;
764

H
Hugh Dickins 已提交
765 766 767
	count = p->swap_map[offset];
	has_cache = count & SWAP_HAS_CACHE;
	count &= ~SWAP_HAS_CACHE;
768

H
Hugh Dickins 已提交
769
	if (usage == SWAP_HAS_CACHE) {
770
		VM_BUG_ON(!has_cache);
H
Hugh Dickins 已提交
771
		has_cache = 0;
H
Hugh Dickins 已提交
772 773 774 775 776 777
	} else if (count == SWAP_MAP_SHMEM) {
		/*
		 * Or we could insist on shmem.c using a special
		 * swap_shmem_free() and free_shmem_swap_and_cache()...
		 */
		count = 0;
H
Hugh Dickins 已提交
778 779 780 781 782 783 784 785 786
	} else if ((count & ~COUNT_CONTINUED) <= SWAP_MAP_MAX) {
		if (count == COUNT_CONTINUED) {
			if (swap_count_continued(p, offset, count))
				count = SWAP_MAP_MAX | COUNT_CONTINUED;
			else
				count = SWAP_MAP_MAX;
		} else
			count--;
	}
H
Hugh Dickins 已提交
787 788 789 790 791 792

	if (!count)
		mem_cgroup_uncharge_swap(entry);

	usage = count | has_cache;
	p->swap_map[offset] = usage;
793 794

	/* free if no reference */
H
Hugh Dickins 已提交
795
	if (!usage) {
796
		dec_cluster_info_page(p, p->cluster_info, offset);
797 798
		if (offset < p->lowest_bit)
			p->lowest_bit = offset;
799 800
		if (offset > p->highest_bit) {
			bool was_full = !p->highest_bit;
801
			p->highest_bit = offset;
802 803 804 805 806 807 808 809 810
			if (was_full && (p->flags & SWP_WRITEOK)) {
				spin_lock(&swap_avail_lock);
				WARN_ON(!plist_node_empty(&p->avail_list));
				if (plist_node_empty(&p->avail_list))
					plist_add(&p->avail_list,
						  &swap_avail_head);
				spin_unlock(&swap_avail_lock);
			}
		}
811
		atomic_long_inc(&nr_swap_pages);
812
		p->inuse_pages--;
813
		frontswap_invalidate_page(p->type, offset);
814 815 816 817 818 819
		if (p->flags & SWP_BLKDEV) {
			struct gendisk *disk = p->bdev->bd_disk;
			if (disk->fops->swap_slot_free_notify)
				disk->fops->swap_slot_free_notify(p->bdev,
								  offset);
		}
L
Linus Torvalds 已提交
820
	}
H
Hugh Dickins 已提交
821 822

	return usage;
L
Linus Torvalds 已提交
823 824 825
}

/*
S
Seth Jennings 已提交
826
 * Caller has made sure that the swap device corresponding to entry
L
Linus Torvalds 已提交
827 828 829 830
 * is still around or has not been recycled.
 */
void swap_free(swp_entry_t entry)
{
831
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
832 833 834

	p = swap_info_get(entry);
	if (p) {
H
Hugh Dickins 已提交
835
		swap_entry_free(p, entry, 1);
836
		spin_unlock(&p->lock);
L
Linus Torvalds 已提交
837 838 839
	}
}

840 841 842
/*
 * Called after dropping swapcache to decrease refcnt to swap entries.
 */
843
void swapcache_free(swp_entry_t entry)
844
{
845 846 847 848
	struct swap_info_struct *p;

	p = swap_info_get(entry);
	if (p) {
849
		swap_entry_free(p, entry, SWAP_HAS_CACHE);
850
		spin_unlock(&p->lock);
851
	}
852 853
}

L
Linus Torvalds 已提交
854
/*
855
 * How many references to page are currently swapped out?
H
Hugh Dickins 已提交
856 857
 * This does not give an exact answer when swap count is continued,
 * but does include the high COUNT_CONTINUED flag to allow for that.
L
Linus Torvalds 已提交
858
 */
859
int page_swapcount(struct page *page)
L
Linus Torvalds 已提交
860
{
861 862
	int count = 0;
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
863 864
	swp_entry_t entry;

H
Hugh Dickins 已提交
865
	entry.val = page_private(page);
L
Linus Torvalds 已提交
866 867
	p = swap_info_get(entry);
	if (p) {
868
		count = swap_count(p->swap_map[swp_offset(entry)]);
869
		spin_unlock(&p->lock);
L
Linus Torvalds 已提交
870
	}
871
	return count;
L
Linus Torvalds 已提交
872 873
}

874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902
/*
 * How many references to @entry are currently swapped out?
 * This considers COUNT_CONTINUED so it returns exact answer.
 */
int swp_swapcount(swp_entry_t entry)
{
	int count, tmp_count, n;
	struct swap_info_struct *p;
	struct page *page;
	pgoff_t offset;
	unsigned char *map;

	p = swap_info_get(entry);
	if (!p)
		return 0;

	count = swap_count(p->swap_map[swp_offset(entry)]);
	if (!(count & COUNT_CONTINUED))
		goto out;

	count &= ~COUNT_CONTINUED;
	n = SWAP_MAP_MAX + 1;

	offset = swp_offset(entry);
	page = vmalloc_to_page(p->swap_map + offset);
	offset &= ~PAGE_MASK;
	VM_BUG_ON(page_private(page) != SWP_CONTINUED);

	do {
903
		page = list_next_entry(page, lru);
904 905 906 907 908 909 910 911 912 913 914 915
		map = kmap_atomic(page);
		tmp_count = map[offset];
		kunmap_atomic(map);

		count += (tmp_count & ~COUNT_CONTINUED) * n;
		n *= (SWAP_CONT_MAX + 1);
	} while (tmp_count & COUNT_CONTINUED);
out:
	spin_unlock(&p->lock);
	return count;
}

L
Linus Torvalds 已提交
916
/*
917 918 919 920
 * We can write to an anon page without COW if there are no other references
 * to it.  And as a side-effect, free up its swap: because the old content
 * on disk will never be read, and seeking back there to write new content
 * later would only waste time away from clustering.
L
Linus Torvalds 已提交
921
 */
922
int reuse_swap_page(struct page *page)
L
Linus Torvalds 已提交
923
{
924 925
	int count;

926
	VM_BUG_ON_PAGE(!PageLocked(page), page);
H
Hugh Dickins 已提交
927 928
	if (unlikely(PageKsm(page)))
		return 0;
929
	count = page_mapcount(page);
930
	if (count <= 1 && PageSwapCache(page)) {
931
		count += page_swapcount(page);
932 933 934 935 936
		if (count == 1 && !PageWriteback(page)) {
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
	}
H
Hugh Dickins 已提交
937
	return count <= 1;
L
Linus Torvalds 已提交
938 939 940
}

/*
941 942
 * If swap is getting full, or if there are no more mappings of this page,
 * then try_to_free_swap is called to free its swap space.
L
Linus Torvalds 已提交
943
 */
944
int try_to_free_swap(struct page *page)
L
Linus Torvalds 已提交
945
{
946
	VM_BUG_ON_PAGE(!PageLocked(page), page);
L
Linus Torvalds 已提交
947 948 949 950 951

	if (!PageSwapCache(page))
		return 0;
	if (PageWriteback(page))
		return 0;
952
	if (page_swapcount(page))
L
Linus Torvalds 已提交
953 954
		return 0;

955 956 957 958 959 960 961 962 963 964 965 966
	/*
	 * Once hibernation has begun to create its image of memory,
	 * there's a danger that one of the calls to try_to_free_swap()
	 * - most probably a call from __try_to_reclaim_swap() while
	 * hibernation is allocating its own swap pages for the image,
	 * but conceivably even a call from memory reclaim - will free
	 * the swap from a page which has already been recorded in the
	 * image as a clean swapcache page, and then reuse its swap for
	 * another page of the image.  On waking from hibernation, the
	 * original page might be freed under memory pressure, then
	 * later read back in from swap, now with the wrong data.
	 *
S
Seth Jennings 已提交
967
	 * Hibernation suspends storage while it is writing the image
968
	 * to disk so check that here.
969
	 */
970
	if (pm_suspended_storage())
971 972
		return 0;

973 974 975
	delete_from_swap_cache(page);
	SetPageDirty(page);
	return 1;
976 977
}

L
Linus Torvalds 已提交
978 979 980 981
/*
 * Free the swap entry like above, but also try to
 * free the page cache entry if it is the last user.
 */
982
int free_swap_and_cache(swp_entry_t entry)
L
Linus Torvalds 已提交
983
{
984
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
985 986
	struct page *page = NULL;

987
	if (non_swap_entry(entry))
988
		return 1;
989

L
Linus Torvalds 已提交
990 991
	p = swap_info_get(entry);
	if (p) {
H
Hugh Dickins 已提交
992
		if (swap_entry_free(p, entry, 1) == SWAP_HAS_CACHE) {
993 994
			page = find_get_page(swap_address_space(entry),
						entry.val);
N
Nick Piggin 已提交
995
			if (page && !trylock_page(page)) {
996 997 998 999
				page_cache_release(page);
				page = NULL;
			}
		}
1000
		spin_unlock(&p->lock);
L
Linus Torvalds 已提交
1001 1002
	}
	if (page) {
1003 1004 1005 1006
		/*
		 * Not mapped elsewhere, or swap space full? Free it!
		 * Also recheck PageSwapCache now page is locked (above).
		 */
1007
		if (PageSwapCache(page) && !PageWriteback(page) &&
1008
				(!page_mapped(page) || vm_swap_full())) {
L
Linus Torvalds 已提交
1009 1010 1011 1012 1013 1014
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
		unlock_page(page);
		page_cache_release(page);
	}
1015
	return p != NULL;
L
Linus Torvalds 已提交
1016 1017
}

1018
#ifdef CONFIG_HIBERNATION
1019
/*
1020
 * Find the swap type that corresponds to given device (if any).
1021
 *
1022 1023 1024 1025
 * @offset - number of the PAGE_SIZE-sized block of the device, starting
 * from 0, in which the swap header is expected to be located.
 *
 * This is needed for the suspend to disk (aka swsusp).
1026
 */
1027
int swap_type_of(dev_t device, sector_t offset, struct block_device **bdev_p)
1028
{
1029
	struct block_device *bdev = NULL;
1030
	int type;
1031

1032 1033 1034
	if (device)
		bdev = bdget(device);

1035
	spin_lock(&swap_lock);
1036 1037
	for (type = 0; type < nr_swapfiles; type++) {
		struct swap_info_struct *sis = swap_info[type];
1038

1039
		if (!(sis->flags & SWP_WRITEOK))
1040
			continue;
1041

1042
		if (!bdev) {
1043
			if (bdev_p)
1044
				*bdev_p = bdgrab(sis->bdev);
1045

1046
			spin_unlock(&swap_lock);
1047
			return type;
1048
		}
1049
		if (bdev == sis->bdev) {
1050
			struct swap_extent *se = &sis->first_swap_extent;
1051 1052

			if (se->start_block == offset) {
1053
				if (bdev_p)
1054
					*bdev_p = bdgrab(sis->bdev);
1055

1056 1057
				spin_unlock(&swap_lock);
				bdput(bdev);
1058
				return type;
1059
			}
1060 1061 1062
		}
	}
	spin_unlock(&swap_lock);
1063 1064 1065
	if (bdev)
		bdput(bdev);

1066 1067 1068
	return -ENODEV;
}

1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080
/*
 * Get the (PAGE_SIZE) block corresponding to given offset on the swapdev
 * corresponding to given index in swap_info (swap type).
 */
sector_t swapdev_block(int type, pgoff_t offset)
{
	struct block_device *bdev;

	if ((unsigned int)type >= nr_swapfiles)
		return 0;
	if (!(swap_info[type]->flags & SWP_WRITEOK))
		return 0;
1081
	return map_swap_entry(swp_entry(type, offset), &bdev);
1082 1083
}

1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
/*
 * Return either the total number of swap pages of given type, or the number
 * of free pages of that type (depending on @free)
 *
 * This is needed for software suspend
 */
unsigned int count_swap_pages(int type, int free)
{
	unsigned int n = 0;

1094 1095 1096 1097
	spin_lock(&swap_lock);
	if ((unsigned int)type < nr_swapfiles) {
		struct swap_info_struct *sis = swap_info[type];

1098
		spin_lock(&sis->lock);
1099 1100
		if (sis->flags & SWP_WRITEOK) {
			n = sis->pages;
1101
			if (free)
1102
				n -= sis->inuse_pages;
1103
		}
1104
		spin_unlock(&sis->lock);
1105
	}
1106
	spin_unlock(&swap_lock);
1107 1108
	return n;
}
1109
#endif /* CONFIG_HIBERNATION */
1110

1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
static inline int maybe_same_pte(pte_t pte, pte_t swp_pte)
{
#ifdef CONFIG_MEM_SOFT_DIRTY
	/*
	 * When pte keeps soft dirty bit the pte generated
	 * from swap entry does not has it, still it's same
	 * pte from logical point of view.
	 */
	pte_t swp_pte_dirty = pte_swp_mksoft_dirty(swp_pte);
	return pte_same(pte, swp_pte) || pte_same(pte, swp_pte_dirty);
#else
	return pte_same(pte, swp_pte);
#endif
}

L
Linus Torvalds 已提交
1126
/*
1127 1128 1129
 * No need to decide whether this PTE shares the swap entry with others,
 * just let do_wp_page work it out if a write is requested later - to
 * force COW, vm_page_prot omits write permission from any private vma.
L
Linus Torvalds 已提交
1130
 */
H
Hugh Dickins 已提交
1131
static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
1132 1133
		unsigned long addr, swp_entry_t entry, struct page *page)
{
1134
	struct page *swapcache;
1135
	struct mem_cgroup *memcg;
H
Hugh Dickins 已提交
1136 1137 1138 1139
	spinlock_t *ptl;
	pte_t *pte;
	int ret = 1;

1140 1141 1142 1143 1144
	swapcache = page;
	page = ksm_might_need_to_copy(page, vma, addr);
	if (unlikely(!page))
		return -ENOMEM;

1145 1146
	if (mem_cgroup_try_charge(page, vma->vm_mm, GFP_KERNEL,
				&memcg, false)) {
H
Hugh Dickins 已提交
1147
		ret = -ENOMEM;
1148 1149
		goto out_nolock;
	}
H
Hugh Dickins 已提交
1150 1151

	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
1152
	if (unlikely(!maybe_same_pte(*pte, swp_entry_to_pte(entry)))) {
1153
		mem_cgroup_cancel_charge(page, memcg, false);
H
Hugh Dickins 已提交
1154 1155 1156
		ret = 0;
		goto out;
	}
1157

K
KAMEZAWA Hiroyuki 已提交
1158
	dec_mm_counter(vma->vm_mm, MM_SWAPENTS);
K
KAMEZAWA Hiroyuki 已提交
1159
	inc_mm_counter(vma->vm_mm, MM_ANONPAGES);
L
Linus Torvalds 已提交
1160 1161 1162
	get_page(page);
	set_pte_at(vma->vm_mm, addr, pte,
		   pte_mkold(mk_pte(page, vma->vm_page_prot)));
1163
	if (page == swapcache) {
1164
		page_add_anon_rmap(page, vma, addr, false);
1165
		mem_cgroup_commit_charge(page, memcg, true, false);
1166
	} else { /* ksm created a completely new copy */
1167
		page_add_new_anon_rmap(page, vma, addr, false);
1168
		mem_cgroup_commit_charge(page, memcg, false, false);
1169 1170
		lru_cache_add_active_or_unevictable(page, vma);
	}
L
Linus Torvalds 已提交
1171 1172 1173 1174 1175 1176
	swap_free(entry);
	/*
	 * Move the page to the active list so it is not
	 * immediately swapped out again after swapon.
	 */
	activate_page(page);
H
Hugh Dickins 已提交
1177 1178
out:
	pte_unmap_unlock(pte, ptl);
1179
out_nolock:
1180 1181 1182 1183
	if (page != swapcache) {
		unlock_page(page);
		put_page(page);
	}
H
Hugh Dickins 已提交
1184
	return ret;
L
Linus Torvalds 已提交
1185 1186 1187 1188 1189 1190 1191
}

static int unuse_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
				unsigned long addr, unsigned long end,
				swp_entry_t entry, struct page *page)
{
	pte_t swp_pte = swp_entry_to_pte(entry);
1192
	pte_t *pte;
1193
	int ret = 0;
L
Linus Torvalds 已提交
1194

H
Hugh Dickins 已提交
1195 1196 1197 1198 1199 1200 1201
	/*
	 * We don't actually need pte lock while scanning for swp_pte: since
	 * we hold page lock and mmap_sem, swp_pte cannot be inserted into the
	 * page table while we're scanning; though it could get zapped, and on
	 * some architectures (e.g. x86_32 with PAE) we might catch a glimpse
	 * of unmatched parts which look like swp_pte, so unuse_pte must
	 * recheck under pte lock.  Scanning without pte lock lets it be
S
Seth Jennings 已提交
1202
	 * preemptable whenever CONFIG_PREEMPT but not CONFIG_HIGHPTE.
H
Hugh Dickins 已提交
1203 1204
	 */
	pte = pte_offset_map(pmd, addr);
L
Linus Torvalds 已提交
1205 1206 1207 1208 1209
	do {
		/*
		 * swapoff spends a _lot_ of time in this loop!
		 * Test inline before going to call unuse_pte.
		 */
1210
		if (unlikely(maybe_same_pte(*pte, swp_pte))) {
H
Hugh Dickins 已提交
1211 1212 1213 1214 1215
			pte_unmap(pte);
			ret = unuse_pte(vma, pmd, addr, entry, page);
			if (ret)
				goto out;
			pte = pte_offset_map(pmd, addr);
L
Linus Torvalds 已提交
1216 1217
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);
H
Hugh Dickins 已提交
1218 1219
	pte_unmap(pte - 1);
out:
1220
	return ret;
L
Linus Torvalds 已提交
1221 1222 1223 1224 1225 1226 1227 1228
}

static inline int unuse_pmd_range(struct vm_area_struct *vma, pud_t *pud,
				unsigned long addr, unsigned long end,
				swp_entry_t entry, struct page *page)
{
	pmd_t *pmd;
	unsigned long next;
1229
	int ret;
L
Linus Torvalds 已提交
1230 1231 1232 1233

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
1234
		if (pmd_none_or_trans_huge_or_clear_bad(pmd))
L
Linus Torvalds 已提交
1235
			continue;
1236 1237 1238
		ret = unuse_pte_range(vma, pmd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
	} while (pmd++, addr = next, addr != end);
	return 0;
}

static inline int unuse_pud_range(struct vm_area_struct *vma, pgd_t *pgd,
				unsigned long addr, unsigned long end,
				swp_entry_t entry, struct page *page)
{
	pud_t *pud;
	unsigned long next;
1249
	int ret;
L
Linus Torvalds 已提交
1250 1251 1252 1253 1254 1255

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
1256 1257 1258
		ret = unuse_pmd_range(vma, pud, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
1259 1260 1261 1262 1263 1264 1265 1266 1267
	} while (pud++, addr = next, addr != end);
	return 0;
}

static int unuse_vma(struct vm_area_struct *vma,
				swp_entry_t entry, struct page *page)
{
	pgd_t *pgd;
	unsigned long addr, end, next;
1268
	int ret;
L
Linus Torvalds 已提交
1269

H
Hugh Dickins 已提交
1270
	if (page_anon_vma(page)) {
L
Linus Torvalds 已提交
1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
		addr = page_address_in_vma(page, vma);
		if (addr == -EFAULT)
			return 0;
		else
			end = addr + PAGE_SIZE;
	} else {
		addr = vma->vm_start;
		end = vma->vm_end;
	}

	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
1286 1287 1288
		ret = unuse_pud_range(vma, pgd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
1289 1290 1291 1292 1293 1294 1295 1296
	} while (pgd++, addr = next, addr != end);
	return 0;
}

static int unuse_mm(struct mm_struct *mm,
				swp_entry_t entry, struct page *page)
{
	struct vm_area_struct *vma;
1297
	int ret = 0;
L
Linus Torvalds 已提交
1298 1299 1300

	if (!down_read_trylock(&mm->mmap_sem)) {
		/*
1301 1302
		 * Activate page so shrink_inactive_list is unlikely to unmap
		 * its ptes while lock is dropped, so swapoff can make progress.
L
Linus Torvalds 已提交
1303
		 */
1304
		activate_page(page);
L
Linus Torvalds 已提交
1305 1306 1307 1308 1309
		unlock_page(page);
		down_read(&mm->mmap_sem);
		lock_page(page);
	}
	for (vma = mm->mmap; vma; vma = vma->vm_next) {
1310
		if (vma->anon_vma && (ret = unuse_vma(vma, entry, page)))
L
Linus Torvalds 已提交
1311 1312 1313
			break;
	}
	up_read(&mm->mmap_sem);
1314
	return (ret < 0)? ret: 0;
L
Linus Torvalds 已提交
1315 1316 1317
}

/*
1318 1319
 * Scan swap_map (or frontswap_map if frontswap parameter is true)
 * from current position to next entry still in use.
L
Linus Torvalds 已提交
1320 1321
 * Recycle to start on reaching the end, returning 0 when empty.
 */
1322
static unsigned int find_next_to_unuse(struct swap_info_struct *si,
1323
					unsigned int prev, bool frontswap)
L
Linus Torvalds 已提交
1324
{
1325 1326
	unsigned int max = si->max;
	unsigned int i = prev;
1327
	unsigned char count;
L
Linus Torvalds 已提交
1328 1329

	/*
1330
	 * No need for swap_lock here: we're just looking
L
Linus Torvalds 已提交
1331 1332
	 * for whether an entry is in use, not modifying it; false
	 * hits are okay, and sys_swapoff() has already prevented new
1333
	 * allocations from this area (while holding swap_lock).
L
Linus Torvalds 已提交
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348
	 */
	for (;;) {
		if (++i >= max) {
			if (!prev) {
				i = 0;
				break;
			}
			/*
			 * No entries in use at top of swap_map,
			 * loop back to start and recheck there.
			 */
			max = prev + 1;
			prev = 0;
			i = 1;
		}
1349 1350 1351 1352 1353 1354
		if (frontswap) {
			if (frontswap_test(si, i))
				break;
			else
				continue;
		}
1355
		count = READ_ONCE(si->swap_map[i]);
1356
		if (count && swap_count(count) != SWAP_MAP_BAD)
L
Linus Torvalds 已提交
1357 1358 1359 1360 1361 1362 1363 1364 1365
			break;
	}
	return i;
}

/*
 * We completely avoid races by reading each swap page in advance,
 * and then search for the process using it.  All the necessary
 * page table adjustments can then be made atomically.
1366 1367 1368
 *
 * if the boolean frontswap is true, only unuse pages_to_unuse pages;
 * pages_to_unuse==0 means all pages; ignored if frontswap is false
L
Linus Torvalds 已提交
1369
 */
1370 1371
int try_to_unuse(unsigned int type, bool frontswap,
		 unsigned long pages_to_unuse)
L
Linus Torvalds 已提交
1372
{
1373
	struct swap_info_struct *si = swap_info[type];
L
Linus Torvalds 已提交
1374
	struct mm_struct *start_mm;
1375 1376 1377 1378 1379
	volatile unsigned char *swap_map; /* swap_map is accessed without
					   * locking. Mark it as volatile
					   * to prevent compiler doing
					   * something odd.
					   */
1380
	unsigned char swcount;
L
Linus Torvalds 已提交
1381 1382
	struct page *page;
	swp_entry_t entry;
1383
	unsigned int i = 0;
L
Linus Torvalds 已提交
1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397
	int retval = 0;

	/*
	 * When searching mms for an entry, a good strategy is to
	 * start at the first mm we freed the previous entry from
	 * (though actually we don't notice whether we or coincidence
	 * freed the entry).  Initialize this start_mm with a hold.
	 *
	 * A simpler strategy would be to start at the last mm we
	 * freed the previous entry from; but that would take less
	 * advantage of mmlist ordering, which clusters forked mms
	 * together, child after parent.  If we race with dup_mmap(), we
	 * prefer to resolve parent before child, lest we miss entries
	 * duplicated after we scanned child: using last mm would invert
H
Hugh Dickins 已提交
1398
	 * that.
L
Linus Torvalds 已提交
1399 1400 1401 1402 1403 1404 1405 1406 1407
	 */
	start_mm = &init_mm;
	atomic_inc(&init_mm.mm_users);

	/*
	 * Keep on scanning until all entries have gone.  Usually,
	 * one pass through swap_map is enough, but not necessarily:
	 * there are races when an instance of an entry might be missed.
	 */
1408
	while ((i = find_next_to_unuse(si, i, frontswap)) != 0) {
L
Linus Torvalds 已提交
1409 1410 1411 1412 1413
		if (signal_pending(current)) {
			retval = -EINTR;
			break;
		}

1414
		/*
L
Linus Torvalds 已提交
1415 1416
		 * Get a page for the entry, using the existing swap
		 * cache page if there is one.  Otherwise, get a clean
1417
		 * page and read the swap into it.
L
Linus Torvalds 已提交
1418 1419 1420
		 */
		swap_map = &si->swap_map[i];
		entry = swp_entry(type, i);
1421 1422
		page = read_swap_cache_async(entry,
					GFP_HIGHUSER_MOVABLE, NULL, 0);
L
Linus Torvalds 已提交
1423 1424 1425 1426 1427 1428 1429
		if (!page) {
			/*
			 * Either swap_duplicate() failed because entry
			 * has been freed independently, and will not be
			 * reused since sys_swapoff() already disabled
			 * allocation from here, or alloc_page() failed.
			 */
1430 1431 1432 1433 1434 1435 1436 1437 1438
			swcount = *swap_map;
			/*
			 * We don't hold lock here, so the swap entry could be
			 * SWAP_MAP_BAD (when the cluster is discarding).
			 * Instead of fail out, We can just skip the swap
			 * entry because swapoff will wait for discarding
			 * finish anyway.
			 */
			if (!swcount || swcount == SWAP_MAP_BAD)
L
Linus Torvalds 已提交
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469
				continue;
			retval = -ENOMEM;
			break;
		}

		/*
		 * Don't hold on to start_mm if it looks like exiting.
		 */
		if (atomic_read(&start_mm->mm_users) == 1) {
			mmput(start_mm);
			start_mm = &init_mm;
			atomic_inc(&init_mm.mm_users);
		}

		/*
		 * Wait for and lock page.  When do_swap_page races with
		 * try_to_unuse, do_swap_page can handle the fault much
		 * faster than try_to_unuse can locate the entry.  This
		 * apparently redundant "wait_on_page_locked" lets try_to_unuse
		 * defer to do_swap_page in such a case - in some tests,
		 * do_swap_page and try_to_unuse repeatedly compete.
		 */
		wait_on_page_locked(page);
		wait_on_page_writeback(page);
		lock_page(page);
		wait_on_page_writeback(page);

		/*
		 * Remove all references to entry.
		 */
		swcount = *swap_map;
H
Hugh Dickins 已提交
1470 1471 1472 1473 1474 1475
		if (swap_count(swcount) == SWAP_MAP_SHMEM) {
			retval = shmem_unuse(entry, page);
			/* page has already been unlocked and released */
			if (retval < 0)
				break;
			continue;
L
Linus Torvalds 已提交
1476
		}
H
Hugh Dickins 已提交
1477 1478 1479
		if (swap_count(swcount) && start_mm != &init_mm)
			retval = unuse_mm(start_mm, entry, page);

1480
		if (swap_count(*swap_map)) {
L
Linus Torvalds 已提交
1481 1482 1483 1484 1485 1486 1487 1488 1489
			int set_start_mm = (*swap_map >= swcount);
			struct list_head *p = &start_mm->mmlist;
			struct mm_struct *new_start_mm = start_mm;
			struct mm_struct *prev_mm = start_mm;
			struct mm_struct *mm;

			atomic_inc(&new_start_mm->mm_users);
			atomic_inc(&prev_mm->mm_users);
			spin_lock(&mmlist_lock);
H
Hugh Dickins 已提交
1490
			while (swap_count(*swap_map) && !retval &&
L
Linus Torvalds 已提交
1491 1492
					(p = p->next) != &start_mm->mmlist) {
				mm = list_entry(p, struct mm_struct, mmlist);
1493
				if (!atomic_inc_not_zero(&mm->mm_users))
L
Linus Torvalds 已提交
1494 1495 1496 1497 1498 1499 1500 1501
					continue;
				spin_unlock(&mmlist_lock);
				mmput(prev_mm);
				prev_mm = mm;

				cond_resched();

				swcount = *swap_map;
1502
				if (!swap_count(swcount)) /* any usage ? */
L
Linus Torvalds 已提交
1503
					;
H
Hugh Dickins 已提交
1504
				else if (mm == &init_mm)
L
Linus Torvalds 已提交
1505
					set_start_mm = 1;
H
Hugh Dickins 已提交
1506
				else
L
Linus Torvalds 已提交
1507
					retval = unuse_mm(mm, entry, page);
1508

1509
				if (set_start_mm && *swap_map < swcount) {
L
Linus Torvalds 已提交
1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539
					mmput(new_start_mm);
					atomic_inc(&mm->mm_users);
					new_start_mm = mm;
					set_start_mm = 0;
				}
				spin_lock(&mmlist_lock);
			}
			spin_unlock(&mmlist_lock);
			mmput(prev_mm);
			mmput(start_mm);
			start_mm = new_start_mm;
		}
		if (retval) {
			unlock_page(page);
			page_cache_release(page);
			break;
		}

		/*
		 * If a reference remains (rare), we would like to leave
		 * the page in the swap cache; but try_to_unmap could
		 * then re-duplicate the entry once we drop page lock,
		 * so we might loop indefinitely; also, that page could
		 * not be swapped out to other storage meanwhile.  So:
		 * delete from cache even if there's another reference,
		 * after ensuring that the data has been saved to disk -
		 * since if the reference remains (rarer), it will be
		 * read from disk into another page.  Splitting into two
		 * pages would be incorrect if swap supported "shared
		 * private" pages, but they are handled by tmpfs files.
H
Hugh Dickins 已提交
1540 1541 1542 1543 1544 1545
		 *
		 * Given how unuse_vma() targets one particular offset
		 * in an anon_vma, once the anon_vma has been determined,
		 * this splitting happens to be just what is needed to
		 * handle where KSM pages have been swapped out: re-reading
		 * is unnecessarily slow, but we can fix that later on.
L
Linus Torvalds 已提交
1546
		 */
1547 1548
		if (swap_count(*swap_map) &&
		     PageDirty(page) && PageSwapCache(page)) {
L
Linus Torvalds 已提交
1549 1550 1551 1552 1553 1554 1555 1556
			struct writeback_control wbc = {
				.sync_mode = WB_SYNC_NONE,
			};

			swap_writepage(page, &wbc);
			lock_page(page);
			wait_on_page_writeback(page);
		}
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566

		/*
		 * It is conceivable that a racing task removed this page from
		 * swap cache just before we acquired the page lock at the top,
		 * or while we dropped it in unuse_mm().  The page might even
		 * be back in swap cache on another swap area: that we must not
		 * delete, since it may not have been written out to swap yet.
		 */
		if (PageSwapCache(page) &&
		    likely(page_private(page) == entry.val))
1567
			delete_from_swap_cache(page);
L
Linus Torvalds 已提交
1568 1569 1570 1571

		/*
		 * So we could skip searching mms once swap count went
		 * to 1, we did not mark any present ptes as dirty: must
1572
		 * mark page dirty so shrink_page_list will preserve it.
L
Linus Torvalds 已提交
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582
		 */
		SetPageDirty(page);
		unlock_page(page);
		page_cache_release(page);

		/*
		 * Make sure that we aren't completely killing
		 * interactive performance.
		 */
		cond_resched();
1583 1584 1585 1586
		if (frontswap && pages_to_unuse > 0) {
			if (!--pages_to_unuse)
				break;
		}
L
Linus Torvalds 已提交
1587 1588 1589 1590 1591 1592 1593
	}

	mmput(start_mm);
	return retval;
}

/*
1594 1595 1596
 * After a successful try_to_unuse, if no swap is now in use, we know
 * we can empty the mmlist.  swap_lock must be held on entry and exit.
 * Note that mmlist_lock nests inside swap_lock, and an mm must be
L
Linus Torvalds 已提交
1597 1598 1599 1600 1601
 * added to the mmlist just after page_duplicate - before would be racy.
 */
static void drain_mmlist(void)
{
	struct list_head *p, *next;
1602
	unsigned int type;
L
Linus Torvalds 已提交
1603

1604 1605
	for (type = 0; type < nr_swapfiles; type++)
		if (swap_info[type]->inuse_pages)
L
Linus Torvalds 已提交
1606 1607 1608 1609 1610 1611 1612 1613 1614
			return;
	spin_lock(&mmlist_lock);
	list_for_each_safe(p, next, &init_mm.mmlist)
		list_del_init(p);
	spin_unlock(&mmlist_lock);
}

/*
 * Use this swapdev's extent info to locate the (PAGE_SIZE) block which
1615 1616 1617
 * corresponds to page offset for the specified swap entry.
 * Note that the type of this function is sector_t, but it returns page offset
 * into the bdev, not sector offset.
L
Linus Torvalds 已提交
1618
 */
1619
static sector_t map_swap_entry(swp_entry_t entry, struct block_device **bdev)
L
Linus Torvalds 已提交
1620
{
H
Hugh Dickins 已提交
1621 1622 1623 1624 1625
	struct swap_info_struct *sis;
	struct swap_extent *start_se;
	struct swap_extent *se;
	pgoff_t offset;

1626
	sis = swap_info[swp_type(entry)];
H
Hugh Dickins 已提交
1627 1628 1629 1630 1631
	*bdev = sis->bdev;

	offset = swp_offset(entry);
	start_se = sis->curr_swap_extent;
	se = start_se;
L
Linus Torvalds 已提交
1632 1633 1634 1635 1636 1637

	for ( ; ; ) {
		if (se->start_page <= offset &&
				offset < (se->start_page + se->nr_pages)) {
			return se->start_block + (offset - se->start_page);
		}
1638
		se = list_next_entry(se, list);
L
Linus Torvalds 已提交
1639 1640 1641 1642 1643
		sis->curr_swap_extent = se;
		BUG_ON(se == start_se);		/* It *must* be present */
	}
}

1644 1645 1646 1647 1648 1649 1650 1651 1652 1653
/*
 * Returns the page offset into bdev for the specified page's swap entry.
 */
sector_t map_swap_page(struct page *page, struct block_device **bdev)
{
	swp_entry_t entry;
	entry.val = page_private(page);
	return map_swap_entry(entry, bdev);
}

L
Linus Torvalds 已提交
1654 1655 1656 1657 1658
/*
 * Free all of a swapdev's extent information
 */
static void destroy_swap_extents(struct swap_info_struct *sis)
{
1659
	while (!list_empty(&sis->first_swap_extent.list)) {
L
Linus Torvalds 已提交
1660 1661
		struct swap_extent *se;

1662
		se = list_first_entry(&sis->first_swap_extent.list,
L
Linus Torvalds 已提交
1663 1664 1665 1666
				struct swap_extent, list);
		list_del(&se->list);
		kfree(se);
	}
1667 1668 1669 1670 1671 1672 1673 1674

	if (sis->flags & SWP_FILE) {
		struct file *swap_file = sis->swap_file;
		struct address_space *mapping = swap_file->f_mapping;

		sis->flags &= ~SWP_FILE;
		mapping->a_ops->swap_deactivate(swap_file);
	}
L
Linus Torvalds 已提交
1675 1676 1677 1678
}

/*
 * Add a block range (and the corresponding page range) into this swapdev's
1679
 * extent list.  The extent list is kept sorted in page order.
L
Linus Torvalds 已提交
1680
 *
1681
 * This function rather assumes that it is called in ascending page order.
L
Linus Torvalds 已提交
1682
 */
1683
int
L
Linus Torvalds 已提交
1684 1685 1686 1687 1688 1689 1690
add_swap_extent(struct swap_info_struct *sis, unsigned long start_page,
		unsigned long nr_pages, sector_t start_block)
{
	struct swap_extent *se;
	struct swap_extent *new_se;
	struct list_head *lh;

1691 1692 1693 1694 1695 1696 1697 1698 1699
	if (start_page == 0) {
		se = &sis->first_swap_extent;
		sis->curr_swap_extent = se;
		se->start_page = 0;
		se->nr_pages = nr_pages;
		se->start_block = start_block;
		return 1;
	} else {
		lh = sis->first_swap_extent.list.prev;	/* Highest extent */
L
Linus Torvalds 已提交
1700
		se = list_entry(lh, struct swap_extent, list);
1701 1702
		BUG_ON(se->start_page + se->nr_pages != start_page);
		if (se->start_block + se->nr_pages == start_block) {
L
Linus Torvalds 已提交
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
			/* Merge it */
			se->nr_pages += nr_pages;
			return 0;
		}
	}

	/*
	 * No merge.  Insert a new extent, preserving ordering.
	 */
	new_se = kmalloc(sizeof(*se), GFP_KERNEL);
	if (new_se == NULL)
		return -ENOMEM;
	new_se->start_page = start_page;
	new_se->nr_pages = nr_pages;
	new_se->start_block = start_block;

1719
	list_add_tail(&new_se->list, &sis->first_swap_extent.list);
1720
	return 1;
L
Linus Torvalds 已提交
1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
}

/*
 * A `swap extent' is a simple thing which maps a contiguous range of pages
 * onto a contiguous range of disk blocks.  An ordered list of swap extents
 * is built at swapon time and is then used at swap_writepage/swap_readpage
 * time for locating where on disk a page belongs.
 *
 * If the swapfile is an S_ISBLK block device, a single extent is installed.
 * This is done so that the main operating code can treat S_ISBLK and S_ISREG
 * swap files identically.
 *
 * Whether the swapdev is an S_ISREG file or an S_ISBLK blockdev, the swap
 * extent list operates in PAGE_SIZE disk blocks.  Both S_ISREG and S_ISBLK
 * swapfiles are handled *identically* after swapon time.
 *
 * For S_ISREG swapfiles, setup_swap_extents() will walk all the file's blocks
 * and will parse them into an ordered extent list, in PAGE_SIZE chunks.  If
 * some stray blocks are found which do not fall within the PAGE_SIZE alignment
 * requirements, they are simply tossed out - we will never use those blocks
 * for swapping.
 *
1743
 * For S_ISREG swapfiles we set S_SWAPFILE across the life of the swapon.  This
L
Linus Torvalds 已提交
1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
 * prevents root from shooting her foot off by ftruncating an in-use swapfile,
 * which will scribble on the fs.
 *
 * The amount of disk space which a single swap extent represents varies.
 * Typically it is in the 1-4 megabyte range.  So we can have hundreds of
 * extents in the list.  To avoid much list walking, we cache the previous
 * search location in `curr_swap_extent', and start new searches from there.
 * This is extremely effective.  The average number of iterations in
 * map_swap_page() has been measured at about 0.3 per page.  - akpm.
 */
1754
static int setup_swap_extents(struct swap_info_struct *sis, sector_t *span)
L
Linus Torvalds 已提交
1755
{
1756 1757 1758
	struct file *swap_file = sis->swap_file;
	struct address_space *mapping = swap_file->f_mapping;
	struct inode *inode = mapping->host;
L
Linus Torvalds 已提交
1759 1760 1761 1762
	int ret;

	if (S_ISBLK(inode->i_mode)) {
		ret = add_swap_extent(sis, 0, sis->max, 0);
1763
		*span = sis->pages;
1764
		return ret;
L
Linus Torvalds 已提交
1765 1766
	}

1767
	if (mapping->a_ops->swap_activate) {
1768
		ret = mapping->a_ops->swap_activate(sis, swap_file, span);
1769 1770 1771 1772 1773
		if (!ret) {
			sis->flags |= SWP_FILE;
			ret = add_swap_extent(sis, 0, sis->max, 0);
			*span = sis->pages;
		}
1774
		return ret;
1775 1776
	}

1777
	return generic_swapfile_activate(sis, swap_file, span);
L
Linus Torvalds 已提交
1778 1779
}

1780
static void _enable_swap_info(struct swap_info_struct *p, int prio,
1781 1782
				unsigned char *swap_map,
				struct swap_cluster_info *cluster_info)
1783 1784 1785 1786 1787
{
	if (prio >= 0)
		p->prio = prio;
	else
		p->prio = --least_priority;
1788 1789 1790 1791 1792 1793
	/*
	 * the plist prio is negated because plist ordering is
	 * low-to-high, while swap ordering is high-to-low
	 */
	p->list.prio = -p->prio;
	p->avail_list.prio = -p->prio;
1794
	p->swap_map = swap_map;
1795
	p->cluster_info = cluster_info;
1796
	p->flags |= SWP_WRITEOK;
1797
	atomic_long_add(p->pages, &nr_swap_pages);
1798 1799
	total_swap_pages += p->pages;

1800 1801
	assert_spin_locked(&swap_lock);
	/*
1802 1803 1804 1805 1806 1807 1808 1809
	 * both lists are plists, and thus priority ordered.
	 * swap_active_head needs to be priority ordered for swapoff(),
	 * which on removal of any swap_info_struct with an auto-assigned
	 * (i.e. negative) priority increments the auto-assigned priority
	 * of any lower-priority swap_info_structs.
	 * swap_avail_head needs to be priority ordered for get_swap_page(),
	 * which allocates swap pages from the highest available priority
	 * swap_info_struct.
1810
	 */
1811 1812 1813 1814
	plist_add(&p->list, &swap_active_head);
	spin_lock(&swap_avail_lock);
	plist_add(&p->avail_list, &swap_avail_head);
	spin_unlock(&swap_avail_lock);
1815 1816 1817 1818
}

static void enable_swap_info(struct swap_info_struct *p, int prio,
				unsigned char *swap_map,
1819
				struct swap_cluster_info *cluster_info,
1820 1821
				unsigned long *frontswap_map)
{
1822
	frontswap_init(p->type, frontswap_map);
1823
	spin_lock(&swap_lock);
1824
	spin_lock(&p->lock);
1825
	 _enable_swap_info(p, prio, swap_map, cluster_info);
1826
	spin_unlock(&p->lock);
1827 1828 1829 1830 1831 1832
	spin_unlock(&swap_lock);
}

static void reinsert_swap_info(struct swap_info_struct *p)
{
	spin_lock(&swap_lock);
1833
	spin_lock(&p->lock);
1834
	_enable_swap_info(p, p->prio, p->swap_map, p->cluster_info);
1835
	spin_unlock(&p->lock);
1836 1837 1838
	spin_unlock(&swap_lock);
}

1839
SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
L
Linus Torvalds 已提交
1840
{
1841
	struct swap_info_struct *p = NULL;
1842
	unsigned char *swap_map;
1843
	struct swap_cluster_info *cluster_info;
1844
	unsigned long *frontswap_map;
L
Linus Torvalds 已提交
1845 1846 1847
	struct file *swap_file, *victim;
	struct address_space *mapping;
	struct inode *inode;
1848
	struct filename *pathname;
1849
	int err, found = 0;
1850
	unsigned int old_block_size;
1851

L
Linus Torvalds 已提交
1852 1853 1854
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

1855 1856
	BUG_ON(!current->mm);

L
Linus Torvalds 已提交
1857 1858
	pathname = getname(specialfile);
	if (IS_ERR(pathname))
X
Xiaotian Feng 已提交
1859
		return PTR_ERR(pathname);
L
Linus Torvalds 已提交
1860

1861
	victim = file_open_name(pathname, O_RDWR|O_LARGEFILE, 0);
L
Linus Torvalds 已提交
1862 1863 1864 1865 1866
	err = PTR_ERR(victim);
	if (IS_ERR(victim))
		goto out;

	mapping = victim->f_mapping;
1867
	spin_lock(&swap_lock);
1868
	plist_for_each_entry(p, &swap_active_head, list) {
H
Hugh Dickins 已提交
1869
		if (p->flags & SWP_WRITEOK) {
1870 1871
			if (p->swap_file->f_mapping == mapping) {
				found = 1;
L
Linus Torvalds 已提交
1872
				break;
1873
			}
L
Linus Torvalds 已提交
1874 1875
		}
	}
1876
	if (!found) {
L
Linus Torvalds 已提交
1877
		err = -EINVAL;
1878
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1879 1880
		goto out_dput;
	}
1881
	if (!security_vm_enough_memory_mm(current->mm, p->pages))
L
Linus Torvalds 已提交
1882 1883 1884
		vm_unacct_memory(p->pages);
	else {
		err = -ENOMEM;
1885
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1886 1887
		goto out_dput;
	}
1888 1889 1890
	spin_lock(&swap_avail_lock);
	plist_del(&p->avail_list, &swap_avail_head);
	spin_unlock(&swap_avail_lock);
1891
	spin_lock(&p->lock);
1892
	if (p->prio < 0) {
1893 1894
		struct swap_info_struct *si = p;

1895
		plist_for_each_entry_continue(si, &swap_active_head, list) {
1896
			si->prio++;
1897 1898
			si->list.prio--;
			si->avail_list.prio--;
1899
		}
1900 1901
		least_priority++;
	}
1902
	plist_del(&p->list, &swap_active_head);
1903
	atomic_long_sub(p->pages, &nr_swap_pages);
L
Linus Torvalds 已提交
1904 1905
	total_swap_pages -= p->pages;
	p->flags &= ~SWP_WRITEOK;
1906
	spin_unlock(&p->lock);
1907
	spin_unlock(&swap_lock);
1908

1909
	set_current_oom_origin();
1910
	err = try_to_unuse(p->type, false, 0); /* force unuse all pages */
1911
	clear_current_oom_origin();
L
Linus Torvalds 已提交
1912 1913 1914

	if (err) {
		/* re-insert swap space back into swap_list */
1915
		reinsert_swap_info(p);
L
Linus Torvalds 已提交
1916 1917
		goto out_dput;
	}
1918

S
Shaohua Li 已提交
1919 1920
	flush_work(&p->discard_work);

1921
	destroy_swap_extents(p);
H
Hugh Dickins 已提交
1922 1923 1924
	if (p->flags & SWP_CONTINUED)
		free_swap_count_continuations(p);

I
Ingo Molnar 已提交
1925
	mutex_lock(&swapon_mutex);
1926
	spin_lock(&swap_lock);
1927
	spin_lock(&p->lock);
1928 1929
	drain_mmlist();

1930 1931 1932
	/* wait for anyone still in scan_swap_map */
	p->highest_bit = 0;		/* cuts scans short */
	while (p->flags >= SWP_SCANNING) {
1933
		spin_unlock(&p->lock);
1934
		spin_unlock(&swap_lock);
1935
		schedule_timeout_uninterruptible(1);
1936
		spin_lock(&swap_lock);
1937
		spin_lock(&p->lock);
1938 1939
	}

L
Linus Torvalds 已提交
1940
	swap_file = p->swap_file;
1941
	old_block_size = p->old_block_size;
L
Linus Torvalds 已提交
1942 1943 1944 1945
	p->swap_file = NULL;
	p->max = 0;
	swap_map = p->swap_map;
	p->swap_map = NULL;
1946 1947
	cluster_info = p->cluster_info;
	p->cluster_info = NULL;
1948
	frontswap_map = frontswap_map_get(p);
1949
	spin_unlock(&p->lock);
1950
	spin_unlock(&swap_lock);
1951
	frontswap_invalidate_area(p->type);
1952
	frontswap_map_set(p, NULL);
I
Ingo Molnar 已提交
1953
	mutex_unlock(&swapon_mutex);
1954 1955
	free_percpu(p->percpu_cluster);
	p->percpu_cluster = NULL;
L
Linus Torvalds 已提交
1956
	vfree(swap_map);
1957
	vfree(cluster_info);
1958
	vfree(frontswap_map);
S
Seth Jennings 已提交
1959
	/* Destroy swap account information */
1960
	swap_cgroup_swapoff(p->type);
1961

L
Linus Torvalds 已提交
1962 1963 1964
	inode = mapping->host;
	if (S_ISBLK(inode->i_mode)) {
		struct block_device *bdev = I_BDEV(inode);
1965
		set_blocksize(bdev, old_block_size);
1966
		blkdev_put(bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
L
Linus Torvalds 已提交
1967
	} else {
1968
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
1969
		inode->i_flags &= ~S_SWAPFILE;
1970
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
1971 1972
	}
	filp_close(swap_file, NULL);
1973 1974 1975 1976 1977 1978 1979 1980 1981 1982

	/*
	 * Clear the SWP_USED flag after all resources are freed so that swapon
	 * can reuse this swap_info in alloc_swap_info() safely.  It is ok to
	 * not hold p->lock after we cleared its SWP_WRITEOK.
	 */
	spin_lock(&swap_lock);
	p->flags = 0;
	spin_unlock(&swap_lock);

L
Linus Torvalds 已提交
1983
	err = 0;
K
Kay Sievers 已提交
1984 1985
	atomic_inc(&proc_poll_event);
	wake_up_interruptible(&proc_poll_wait);
L
Linus Torvalds 已提交
1986 1987 1988 1989

out_dput:
	filp_close(victim, NULL);
out:
X
Xiaotian Feng 已提交
1990
	putname(pathname);
L
Linus Torvalds 已提交
1991 1992 1993 1994
	return err;
}

#ifdef CONFIG_PROC_FS
K
Kay Sievers 已提交
1995 1996
static unsigned swaps_poll(struct file *file, poll_table *wait)
{
1997
	struct seq_file *seq = file->private_data;
K
Kay Sievers 已提交
1998 1999 2000

	poll_wait(file, &proc_poll_wait, wait);

2001 2002
	if (seq->poll_event != atomic_read(&proc_poll_event)) {
		seq->poll_event = atomic_read(&proc_poll_event);
K
Kay Sievers 已提交
2003 2004 2005 2006 2007 2008
		return POLLIN | POLLRDNORM | POLLERR | POLLPRI;
	}

	return POLLIN | POLLRDNORM;
}

L
Linus Torvalds 已提交
2009 2010 2011
/* iterator */
static void *swap_start(struct seq_file *swap, loff_t *pos)
{
2012 2013
	struct swap_info_struct *si;
	int type;
L
Linus Torvalds 已提交
2014 2015
	loff_t l = *pos;

I
Ingo Molnar 已提交
2016
	mutex_lock(&swapon_mutex);
L
Linus Torvalds 已提交
2017

2018 2019 2020
	if (!l)
		return SEQ_START_TOKEN;

2021 2022 2023 2024
	for (type = 0; type < nr_swapfiles; type++) {
		smp_rmb();	/* read nr_swapfiles before swap_info[type] */
		si = swap_info[type];
		if (!(si->flags & SWP_USED) || !si->swap_map)
L
Linus Torvalds 已提交
2025
			continue;
2026
		if (!--l)
2027
			return si;
L
Linus Torvalds 已提交
2028 2029 2030 2031 2032 2033 2034
	}

	return NULL;
}

static void *swap_next(struct seq_file *swap, void *v, loff_t *pos)
{
2035 2036
	struct swap_info_struct *si = v;
	int type;
L
Linus Torvalds 已提交
2037

2038
	if (v == SEQ_START_TOKEN)
2039 2040 2041
		type = 0;
	else
		type = si->type + 1;
2042

2043 2044 2045 2046
	for (; type < nr_swapfiles; type++) {
		smp_rmb();	/* read nr_swapfiles before swap_info[type] */
		si = swap_info[type];
		if (!(si->flags & SWP_USED) || !si->swap_map)
L
Linus Torvalds 已提交
2047 2048
			continue;
		++*pos;
2049
		return si;
L
Linus Torvalds 已提交
2050 2051 2052 2053 2054 2055 2056
	}

	return NULL;
}

static void swap_stop(struct seq_file *swap, void *v)
{
I
Ingo Molnar 已提交
2057
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
2058 2059 2060 2061
}

static int swap_show(struct seq_file *swap, void *v)
{
2062
	struct swap_info_struct *si = v;
L
Linus Torvalds 已提交
2063 2064 2065
	struct file *file;
	int len;

2066
	if (si == SEQ_START_TOKEN) {
2067 2068 2069
		seq_puts(swap,"Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
		return 0;
	}
L
Linus Torvalds 已提交
2070

2071
	file = si->swap_file;
M
Miklos Szeredi 已提交
2072
	len = seq_file_path(swap, file, " \t\n\\");
2073
	seq_printf(swap, "%*s%s\t%u\t%u\t%d\n",
2074
			len < 40 ? 40 - len : 1, " ",
A
Al Viro 已提交
2075
			S_ISBLK(file_inode(file)->i_mode) ?
L
Linus Torvalds 已提交
2076
				"partition" : "file\t",
2077 2078 2079
			si->pages << (PAGE_SHIFT - 10),
			si->inuse_pages << (PAGE_SHIFT - 10),
			si->prio);
L
Linus Torvalds 已提交
2080 2081 2082
	return 0;
}

2083
static const struct seq_operations swaps_op = {
L
Linus Torvalds 已提交
2084 2085 2086 2087 2088 2089 2090 2091
	.start =	swap_start,
	.next =		swap_next,
	.stop =		swap_stop,
	.show =		swap_show
};

static int swaps_open(struct inode *inode, struct file *file)
{
2092
	struct seq_file *seq;
K
Kay Sievers 已提交
2093 2094 2095
	int ret;

	ret = seq_open(file, &swaps_op);
2096
	if (ret)
K
Kay Sievers 已提交
2097 2098
		return ret;

2099 2100 2101
	seq = file->private_data;
	seq->poll_event = atomic_read(&proc_poll_event);
	return 0;
L
Linus Torvalds 已提交
2102 2103
}

2104
static const struct file_operations proc_swaps_operations = {
L
Linus Torvalds 已提交
2105 2106 2107 2108
	.open		= swaps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= seq_release,
K
Kay Sievers 已提交
2109
	.poll		= swaps_poll,
L
Linus Torvalds 已提交
2110 2111 2112 2113
};

static int __init procswaps_init(void)
{
2114
	proc_create("swaps", 0, NULL, &proc_swaps_operations);
L
Linus Torvalds 已提交
2115 2116 2117 2118 2119
	return 0;
}
__initcall(procswaps_init);
#endif /* CONFIG_PROC_FS */

J
Jan Beulich 已提交
2120 2121 2122 2123 2124 2125 2126 2127 2128
#ifdef MAX_SWAPFILES_CHECK
static int __init max_swapfiles_check(void)
{
	MAX_SWAPFILES_CHECK();
	return 0;
}
late_initcall(max_swapfiles_check);
#endif

2129
static struct swap_info_struct *alloc_swap_info(void)
L
Linus Torvalds 已提交
2130
{
2131
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
2132
	unsigned int type;
2133 2134 2135

	p = kzalloc(sizeof(*p), GFP_KERNEL);
	if (!p)
2136
		return ERR_PTR(-ENOMEM);
2137

2138
	spin_lock(&swap_lock);
2139 2140
	for (type = 0; type < nr_swapfiles; type++) {
		if (!(swap_info[type]->flags & SWP_USED))
L
Linus Torvalds 已提交
2141
			break;
2142
	}
2143
	if (type >= MAX_SWAPFILES) {
2144
		spin_unlock(&swap_lock);
2145
		kfree(p);
2146
		return ERR_PTR(-EPERM);
L
Linus Torvalds 已提交
2147
	}
2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165
	if (type >= nr_swapfiles) {
		p->type = type;
		swap_info[type] = p;
		/*
		 * Write swap_info[type] before nr_swapfiles, in case a
		 * racing procfs swap_start() or swap_next() is reading them.
		 * (We never shrink nr_swapfiles, we never free this entry.)
		 */
		smp_wmb();
		nr_swapfiles++;
	} else {
		kfree(p);
		p = swap_info[type];
		/*
		 * Do not memset this entry: a racing procfs swap_next()
		 * would be relying on p->type to remain valid.
		 */
	}
2166
	INIT_LIST_HEAD(&p->first_swap_extent.list);
2167 2168
	plist_node_init(&p->list, 0);
	plist_node_init(&p->avail_list, 0);
L
Linus Torvalds 已提交
2169
	p->flags = SWP_USED;
2170
	spin_unlock(&swap_lock);
2171
	spin_lock_init(&p->lock);
2172

2173 2174 2175
	return p;
}

2176 2177 2178 2179 2180 2181 2182
static int claim_swapfile(struct swap_info_struct *p, struct inode *inode)
{
	int error;

	if (S_ISBLK(inode->i_mode)) {
		p->bdev = bdgrab(I_BDEV(inode));
		error = blkdev_get(p->bdev,
2183
				   FMODE_READ | FMODE_WRITE | FMODE_EXCL, p);
2184 2185
		if (error < 0) {
			p->bdev = NULL;
2186
			return error;
2187 2188 2189 2190
		}
		p->old_block_size = block_size(p->bdev);
		error = set_blocksize(p->bdev, PAGE_SIZE);
		if (error < 0)
2191
			return error;
2192 2193 2194 2195
		p->flags |= SWP_BLKDEV;
	} else if (S_ISREG(inode->i_mode)) {
		p->bdev = inode->i_sb->s_bdev;
		mutex_lock(&inode->i_mutex);
2196 2197 2198 2199
		if (IS_SWAPFILE(inode))
			return -EBUSY;
	} else
		return -EINVAL;
2200 2201 2202 2203

	return 0;
}

2204 2205 2206 2207 2208 2209 2210
static unsigned long read_swap_header(struct swap_info_struct *p,
					union swap_header *swap_header,
					struct inode *inode)
{
	int i;
	unsigned long maxpages;
	unsigned long swapfilepages;
2211
	unsigned long last_page;
2212 2213

	if (memcmp("SWAPSPACE2", swap_header->magic.magic, 10)) {
2214
		pr_err("Unable to find swap-space signature\n");
2215
		return 0;
2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227
	}

	/* swap partition endianess hack... */
	if (swab32(swap_header->info.version) == 1) {
		swab32s(&swap_header->info.version);
		swab32s(&swap_header->info.last_page);
		swab32s(&swap_header->info.nr_badpages);
		for (i = 0; i < swap_header->info.nr_badpages; i++)
			swab32s(&swap_header->info.badpages[i]);
	}
	/* Check the swap header's sub-version */
	if (swap_header->info.version != 1) {
2228 2229
		pr_warn("Unable to handle swap header version %d\n",
			swap_header->info.version);
2230
		return 0;
2231 2232 2233 2234 2235 2236 2237 2238
	}

	p->lowest_bit  = 1;
	p->cluster_next = 1;
	p->cluster_nr = 0;

	/*
	 * Find out how many pages are allowed for a single swap
2239
	 * device. There are two limiting factors: 1) the number
2240 2241
	 * of bits for the swap offset in the swp_entry_t type, and
	 * 2) the number of bits in the swap pte as defined by the
2242
	 * different architectures. In order to find the
2243
	 * largest possible bit mask, a swap entry with swap type 0
2244
	 * and swap offset ~0UL is created, encoded to a swap pte,
2245
	 * decoded to a swp_entry_t again, and finally the swap
2246 2247 2248
	 * offset is extracted. This will mask all the bits from
	 * the initial ~0UL mask that can't be encoded in either
	 * the swp_entry_t or the architecture definition of a
2249
	 * swap pte.
2250 2251
	 */
	maxpages = swp_offset(pte_to_swp_entry(
2252
			swp_entry_to_pte(swp_entry(0, ~0UL)))) + 1;
2253 2254
	last_page = swap_header->info.last_page;
	if (last_page > maxpages) {
2255
		pr_warn("Truncating oversized swap area, only using %luk out of %luk\n",
2256 2257 2258 2259 2260
			maxpages << (PAGE_SHIFT - 10),
			last_page << (PAGE_SHIFT - 10));
	}
	if (maxpages > last_page) {
		maxpages = last_page + 1;
2261 2262 2263 2264 2265 2266 2267
		/* p->max is an unsigned int: don't overflow it */
		if ((unsigned int)maxpages == 0)
			maxpages = UINT_MAX;
	}
	p->highest_bit = maxpages - 1;

	if (!maxpages)
2268
		return 0;
2269 2270
	swapfilepages = i_size_read(inode) >> PAGE_SHIFT;
	if (swapfilepages && maxpages > swapfilepages) {
2271
		pr_warn("Swap area shorter than signature indicates\n");
2272
		return 0;
2273 2274
	}
	if (swap_header->info.nr_badpages && S_ISREG(inode->i_mode))
2275
		return 0;
2276
	if (swap_header->info.nr_badpages > MAX_SWAP_BADPAGES)
2277
		return 0;
2278 2279 2280 2281

	return maxpages;
}

2282 2283 2284
static int setup_swap_map_and_extents(struct swap_info_struct *p,
					union swap_header *swap_header,
					unsigned char *swap_map,
2285
					struct swap_cluster_info *cluster_info,
2286 2287 2288 2289 2290 2291
					unsigned long maxpages,
					sector_t *span)
{
	int i;
	unsigned int nr_good_pages;
	int nr_extents;
2292 2293
	unsigned long nr_clusters = DIV_ROUND_UP(maxpages, SWAPFILE_CLUSTER);
	unsigned long idx = p->cluster_next / SWAPFILE_CLUSTER;
2294 2295 2296

	nr_good_pages = maxpages - 1;	/* omit header page */

2297 2298
	cluster_set_null(&p->free_cluster_head);
	cluster_set_null(&p->free_cluster_tail);
S
Shaohua Li 已提交
2299 2300
	cluster_set_null(&p->discard_cluster_head);
	cluster_set_null(&p->discard_cluster_tail);
2301

2302 2303
	for (i = 0; i < swap_header->info.nr_badpages; i++) {
		unsigned int page_nr = swap_header->info.badpages[i];
2304 2305
		if (page_nr == 0 || page_nr > swap_header->info.last_page)
			return -EINVAL;
2306 2307 2308
		if (page_nr < maxpages) {
			swap_map[page_nr] = SWAP_MAP_BAD;
			nr_good_pages--;
2309 2310 2311 2312 2313
			/*
			 * Haven't marked the cluster free yet, no list
			 * operation involved
			 */
			inc_cluster_info_page(p, cluster_info, page_nr);
2314 2315 2316
		}
	}

2317 2318 2319 2320
	/* Haven't marked the cluster free yet, no list operation involved */
	for (i = maxpages; i < round_up(maxpages, SWAPFILE_CLUSTER); i++)
		inc_cluster_info_page(p, cluster_info, i);

2321 2322
	if (nr_good_pages) {
		swap_map[0] = SWAP_MAP_BAD;
2323 2324 2325 2326 2327
		/*
		 * Not mark the cluster free yet, no list
		 * operation involved
		 */
		inc_cluster_info_page(p, cluster_info, 0);
2328 2329 2330
		p->max = maxpages;
		p->pages = nr_good_pages;
		nr_extents = setup_swap_extents(p, span);
2331 2332
		if (nr_extents < 0)
			return nr_extents;
2333 2334 2335
		nr_good_pages = p->pages;
	}
	if (!nr_good_pages) {
2336
		pr_warn("Empty swap-file\n");
2337
		return -EINVAL;
2338 2339
	}

2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363
	if (!cluster_info)
		return nr_extents;

	for (i = 0; i < nr_clusters; i++) {
		if (!cluster_count(&cluster_info[idx])) {
			cluster_set_flag(&cluster_info[idx], CLUSTER_FLAG_FREE);
			if (cluster_is_null(&p->free_cluster_head)) {
				cluster_set_next_flag(&p->free_cluster_head,
								idx, 0);
				cluster_set_next_flag(&p->free_cluster_tail,
								idx, 0);
			} else {
				unsigned int tail;

				tail = cluster_next(&p->free_cluster_tail);
				cluster_set_next(&cluster_info[tail], idx);
				cluster_set_next_flag(&p->free_cluster_tail,
								idx, 0);
			}
		}
		idx++;
		if (idx == nr_clusters)
			idx = 0;
	}
2364 2365 2366
	return nr_extents;
}

2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380
/*
 * Helper to sys_swapon determining if a given swap
 * backing device queue supports DISCARD operations.
 */
static bool swap_discardable(struct swap_info_struct *si)
{
	struct request_queue *q = bdev_get_queue(si->bdev);

	if (!q || !blk_queue_discard(q))
		return false;

	return true;
}

2381 2382 2383
SYSCALL_DEFINE2(swapon, const char __user *, specialfile, int, swap_flags)
{
	struct swap_info_struct *p;
2384
	struct filename *name;
2385 2386
	struct file *swap_file = NULL;
	struct address_space *mapping;
2387
	int prio;
2388 2389
	int error;
	union swap_header *swap_header;
2390
	int nr_extents;
2391 2392 2393
	sector_t span;
	unsigned long maxpages;
	unsigned char *swap_map = NULL;
2394
	struct swap_cluster_info *cluster_info = NULL;
2395
	unsigned long *frontswap_map = NULL;
2396 2397 2398
	struct page *page = NULL;
	struct inode *inode = NULL;

2399 2400 2401
	if (swap_flags & ~SWAP_FLAGS_VALID)
		return -EINVAL;

2402 2403 2404 2405
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	p = alloc_swap_info();
2406 2407
	if (IS_ERR(p))
		return PTR_ERR(p);
2408

S
Shaohua Li 已提交
2409 2410
	INIT_WORK(&p->discard_work, swap_discard_work);

L
Linus Torvalds 已提交
2411 2412
	name = getname(specialfile);
	if (IS_ERR(name)) {
2413
		error = PTR_ERR(name);
L
Linus Torvalds 已提交
2414
		name = NULL;
2415
		goto bad_swap;
L
Linus Torvalds 已提交
2416
	}
2417
	swap_file = file_open_name(name, O_RDWR|O_LARGEFILE, 0);
L
Linus Torvalds 已提交
2418
	if (IS_ERR(swap_file)) {
2419
		error = PTR_ERR(swap_file);
L
Linus Torvalds 已提交
2420
		swap_file = NULL;
2421
		goto bad_swap;
L
Linus Torvalds 已提交
2422 2423 2424 2425
	}

	p->swap_file = swap_file;
	mapping = swap_file->f_mapping;
2426
	inode = mapping->host;
2427

2428
	/* If S_ISREG(inode->i_mode) will do mutex_lock(&inode->i_mutex); */
2429 2430
	error = claim_swapfile(p, inode);
	if (unlikely(error))
L
Linus Torvalds 已提交
2431 2432 2433 2434 2435 2436 2437 2438 2439
		goto bad_swap;

	/*
	 * Read the swap header.
	 */
	if (!mapping->a_ops->readpage) {
		error = -EINVAL;
		goto bad_swap;
	}
2440
	page = read_mapping_page(mapping, 0, swap_file);
L
Linus Torvalds 已提交
2441 2442 2443 2444
	if (IS_ERR(page)) {
		error = PTR_ERR(page);
		goto bad_swap;
	}
2445
	swap_header = kmap(page);
L
Linus Torvalds 已提交
2446

2447 2448
	maxpages = read_swap_header(p, swap_header, inode);
	if (unlikely(!maxpages)) {
L
Linus Torvalds 已提交
2449 2450 2451
		error = -EINVAL;
		goto bad_swap;
	}
2452

2453
	/* OK, set up the swap map and apply the bad block list */
2454
	swap_map = vzalloc(maxpages);
2455 2456 2457 2458
	if (!swap_map) {
		error = -ENOMEM;
		goto bad_swap;
	}
2459
	if (p->bdev && blk_queue_nonrot(bdev_get_queue(p->bdev))) {
2460 2461
		int cpu;

2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474
		p->flags |= SWP_SOLIDSTATE;
		/*
		 * select a random position to start with to help wear leveling
		 * SSD
		 */
		p->cluster_next = 1 + (prandom_u32() % p->highest_bit);

		cluster_info = vzalloc(DIV_ROUND_UP(maxpages,
			SWAPFILE_CLUSTER) * sizeof(*cluster_info));
		if (!cluster_info) {
			error = -ENOMEM;
			goto bad_swap;
		}
2475 2476 2477 2478 2479
		p->percpu_cluster = alloc_percpu(struct percpu_cluster);
		if (!p->percpu_cluster) {
			error = -ENOMEM;
			goto bad_swap;
		}
2480
		for_each_possible_cpu(cpu) {
2481
			struct percpu_cluster *cluster;
2482
			cluster = per_cpu_ptr(p->percpu_cluster, cpu);
2483 2484
			cluster_set_null(&cluster->index);
		}
2485
	}
L
Linus Torvalds 已提交
2486

2487 2488 2489 2490
	error = swap_cgroup_swapon(p->type, maxpages);
	if (error)
		goto bad_swap;

2491
	nr_extents = setup_swap_map_and_extents(p, swap_header, swap_map,
2492
		cluster_info, maxpages, &span);
2493 2494
	if (unlikely(nr_extents < 0)) {
		error = nr_extents;
L
Linus Torvalds 已提交
2495 2496
		goto bad_swap;
	}
2497 2498
	/* frontswap enabled? set up bit-per-page map for frontswap */
	if (frontswap_enabled)
2499
		frontswap_map = vzalloc(BITS_TO_LONGS(maxpages) * sizeof(long));
L
Linus Torvalds 已提交
2500

2501 2502 2503 2504 2505 2506 2507 2508 2509
	if (p->bdev &&(swap_flags & SWAP_FLAG_DISCARD) && swap_discardable(p)) {
		/*
		 * When discard is enabled for swap with no particular
		 * policy flagged, we set all swap discard flags here in
		 * order to sustain backward compatibility with older
		 * swapon(8) releases.
		 */
		p->flags |= (SWP_DISCARDABLE | SWP_AREA_DISCARD |
			     SWP_PAGE_DISCARD);
2510

2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527
		/*
		 * By flagging sys_swapon, a sysadmin can tell us to
		 * either do single-time area discards only, or to just
		 * perform discards for released swap page-clusters.
		 * Now it's time to adjust the p->flags accordingly.
		 */
		if (swap_flags & SWAP_FLAG_DISCARD_ONCE)
			p->flags &= ~SWP_PAGE_DISCARD;
		else if (swap_flags & SWAP_FLAG_DISCARD_PAGES)
			p->flags &= ~SWP_AREA_DISCARD;

		/* issue a swapon-time discard if it's still required */
		if (p->flags & SWP_AREA_DISCARD) {
			int err = discard_swap(p);
			if (unlikely(err))
				pr_err("swapon: discard_swap(%p): %d\n",
					p, err);
2528
		}
2529
	}
2530

I
Ingo Molnar 已提交
2531
	mutex_lock(&swapon_mutex);
2532
	prio = -1;
2533
	if (swap_flags & SWAP_FLAG_PREFER)
2534
		prio =
2535
		  (swap_flags & SWAP_FLAG_PRIO_MASK) >> SWAP_FLAG_PRIO_SHIFT;
2536
	enable_swap_info(p, prio, swap_map, cluster_info, frontswap_map);
2537

2538
	pr_info("Adding %uk swap on %s.  "
2539
			"Priority:%d extents:%d across:%lluk %s%s%s%s%s\n",
2540
		p->pages<<(PAGE_SHIFT-10), name->name, p->prio,
2541 2542
		nr_extents, (unsigned long long)span<<(PAGE_SHIFT-10),
		(p->flags & SWP_SOLIDSTATE) ? "SS" : "",
2543
		(p->flags & SWP_DISCARDABLE) ? "D" : "",
2544 2545
		(p->flags & SWP_AREA_DISCARD) ? "s" : "",
		(p->flags & SWP_PAGE_DISCARD) ? "c" : "",
2546
		(frontswap_map) ? "FS" : "");
2547

I
Ingo Molnar 已提交
2548
	mutex_unlock(&swapon_mutex);
K
Kay Sievers 已提交
2549 2550 2551
	atomic_inc(&proc_poll_event);
	wake_up_interruptible(&proc_poll_wait);

2552 2553
	if (S_ISREG(inode->i_mode))
		inode->i_flags |= S_SWAPFILE;
L
Linus Torvalds 已提交
2554 2555 2556
	error = 0;
	goto out;
bad_swap:
2557 2558
	free_percpu(p->percpu_cluster);
	p->percpu_cluster = NULL;
2559
	if (inode && S_ISBLK(inode->i_mode) && p->bdev) {
2560 2561
		set_blocksize(p->bdev, p->old_block_size);
		blkdev_put(p->bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
L
Linus Torvalds 已提交
2562
	}
2563
	destroy_swap_extents(p);
2564
	swap_cgroup_swapoff(p->type);
2565
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
2566 2567
	p->swap_file = NULL;
	p->flags = 0;
2568
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2569
	vfree(swap_map);
2570
	vfree(cluster_info);
2571
	if (swap_file) {
2572
		if (inode && S_ISREG(inode->i_mode)) {
2573
			mutex_unlock(&inode->i_mutex);
2574 2575
			inode = NULL;
		}
L
Linus Torvalds 已提交
2576
		filp_close(swap_file, NULL);
2577
	}
L
Linus Torvalds 已提交
2578 2579 2580 2581 2582 2583 2584
out:
	if (page && !IS_ERR(page)) {
		kunmap(page);
		page_cache_release(page);
	}
	if (name)
		putname(name);
2585
	if (inode && S_ISREG(inode->i_mode))
2586
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
2587 2588 2589 2590 2591
	return error;
}

void si_swapinfo(struct sysinfo *val)
{
2592
	unsigned int type;
L
Linus Torvalds 已提交
2593 2594
	unsigned long nr_to_be_unused = 0;

2595
	spin_lock(&swap_lock);
2596 2597 2598 2599 2600
	for (type = 0; type < nr_swapfiles; type++) {
		struct swap_info_struct *si = swap_info[type];

		if ((si->flags & SWP_USED) && !(si->flags & SWP_WRITEOK))
			nr_to_be_unused += si->inuse_pages;
L
Linus Torvalds 已提交
2601
	}
2602
	val->freeswap = atomic_long_read(&nr_swap_pages) + nr_to_be_unused;
L
Linus Torvalds 已提交
2603
	val->totalswap = total_swap_pages + nr_to_be_unused;
2604
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2605 2606 2607 2608 2609
}

/*
 * Verify that a swap entry is valid and increment its swap map count.
 *
2610 2611 2612 2613 2614 2615
 * Returns error code in following case.
 * - success -> 0
 * - swp_entry is invalid -> EINVAL
 * - swp_entry is migration entry -> EINVAL
 * - swap-cache reference is requested but there is already one. -> EEXIST
 * - swap-cache reference is requested but the entry is not used. -> ENOENT
H
Hugh Dickins 已提交
2616
 * - swap-mapped reference requested but needs continued swap count. -> ENOMEM
L
Linus Torvalds 已提交
2617
 */
2618
static int __swap_duplicate(swp_entry_t entry, unsigned char usage)
L
Linus Torvalds 已提交
2619
{
2620
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
2621
	unsigned long offset, type;
2622 2623
	unsigned char count;
	unsigned char has_cache;
H
Hugh Dickins 已提交
2624
	int err = -EINVAL;
L
Linus Torvalds 已提交
2625

2626
	if (non_swap_entry(entry))
H
Hugh Dickins 已提交
2627
		goto out;
2628

L
Linus Torvalds 已提交
2629 2630 2631
	type = swp_type(entry);
	if (type >= nr_swapfiles)
		goto bad_file;
2632
	p = swap_info[type];
L
Linus Torvalds 已提交
2633 2634
	offset = swp_offset(entry);

2635
	spin_lock(&p->lock);
2636 2637 2638
	if (unlikely(offset >= p->max))
		goto unlock_out;

H
Hugh Dickins 已提交
2639
	count = p->swap_map[offset];
2640 2641 2642 2643 2644 2645 2646 2647 2648 2649

	/*
	 * swapin_readahead() doesn't check if a swap entry is valid, so the
	 * swap entry could be SWAP_MAP_BAD. Check here with lock held.
	 */
	if (unlikely(swap_count(count) == SWAP_MAP_BAD)) {
		err = -ENOENT;
		goto unlock_out;
	}

H
Hugh Dickins 已提交
2650 2651 2652
	has_cache = count & SWAP_HAS_CACHE;
	count &= ~SWAP_HAS_CACHE;
	err = 0;
2653

H
Hugh Dickins 已提交
2654
	if (usage == SWAP_HAS_CACHE) {
2655 2656

		/* set SWAP_HAS_CACHE if there is no cache and entry is used */
H
Hugh Dickins 已提交
2657 2658 2659 2660 2661 2662
		if (!has_cache && count)
			has_cache = SWAP_HAS_CACHE;
		else if (has_cache)		/* someone else added cache */
			err = -EEXIST;
		else				/* no users remaining */
			err = -ENOENT;
2663 2664

	} else if (count || has_cache) {
H
Hugh Dickins 已提交
2665

H
Hugh Dickins 已提交
2666 2667 2668
		if ((count & ~COUNT_CONTINUED) < SWAP_MAP_MAX)
			count += usage;
		else if ((count & ~COUNT_CONTINUED) > SWAP_MAP_MAX)
H
Hugh Dickins 已提交
2669
			err = -EINVAL;
H
Hugh Dickins 已提交
2670 2671 2672 2673
		else if (swap_count_continued(p, offset, count))
			count = COUNT_CONTINUED;
		else
			err = -ENOMEM;
2674
	} else
H
Hugh Dickins 已提交
2675 2676 2677 2678
		err = -ENOENT;			/* unused swap entry */

	p->swap_map[offset] = count | has_cache;

2679
unlock_out:
2680
	spin_unlock(&p->lock);
L
Linus Torvalds 已提交
2681
out:
H
Hugh Dickins 已提交
2682
	return err;
L
Linus Torvalds 已提交
2683 2684

bad_file:
2685
	pr_err("swap_dup: %s%08lx\n", Bad_file, entry.val);
L
Linus Torvalds 已提交
2686 2687
	goto out;
}
H
Hugh Dickins 已提交
2688

H
Hugh Dickins 已提交
2689 2690 2691 2692 2693 2694 2695 2696 2697
/*
 * Help swapoff by noting that swap entry belongs to shmem/tmpfs
 * (in which case its reference count is never incremented).
 */
void swap_shmem_alloc(swp_entry_t entry)
{
	__swap_duplicate(entry, SWAP_MAP_SHMEM);
}

2698
/*
2699 2700 2701 2702 2703
 * Increase reference count of swap entry by 1.
 * Returns 0 for success, or -ENOMEM if a swap_count_continuation is required
 * but could not be atomically allocated.  Returns 0, just as if it succeeded,
 * if __swap_duplicate() fails for another reason (-EINVAL or -ENOENT), which
 * might occur if a page table entry has got corrupted.
2704
 */
H
Hugh Dickins 已提交
2705
int swap_duplicate(swp_entry_t entry)
2706
{
H
Hugh Dickins 已提交
2707 2708 2709 2710 2711
	int err = 0;

	while (!err && __swap_duplicate(entry, 1) == -ENOMEM)
		err = add_swap_count_continuation(entry, GFP_ATOMIC);
	return err;
2712
}
L
Linus Torvalds 已提交
2713

2714
/*
2715 2716
 * @entry: swap entry for which we allocate swap cache.
 *
2717
 * Called when allocating swap cache for existing swap entry,
2718 2719 2720
 * This can return error codes. Returns 0 at success.
 * -EBUSY means there is a swap cache.
 * Note: return code is different from swap_duplicate().
2721 2722 2723
 */
int swapcache_prepare(swp_entry_t entry)
{
H
Hugh Dickins 已提交
2724
	return __swap_duplicate(entry, SWAP_HAS_CACHE);
2725 2726
}

2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738
struct swap_info_struct *page_swap_info(struct page *page)
{
	swp_entry_t swap = { .val = page_private(page) };
	BUG_ON(!PageSwapCache(page));
	return swap_info[swp_type(swap)];
}

/*
 * out-of-line __page_file_ methods to avoid include hell.
 */
struct address_space *__page_file_mapping(struct page *page)
{
2739
	VM_BUG_ON_PAGE(!PageSwapCache(page), page);
2740 2741 2742 2743 2744 2745 2746
	return page_swap_info(page)->swap_file->f_mapping;
}
EXPORT_SYMBOL_GPL(__page_file_mapping);

pgoff_t __page_file_index(struct page *page)
{
	swp_entry_t swap = { .val = page_private(page) };
2747
	VM_BUG_ON_PAGE(!PageSwapCache(page), page);
2748 2749 2750 2751
	return swp_offset(swap);
}
EXPORT_SYMBOL_GPL(__page_file_index);

H
Hugh Dickins 已提交
2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804
/*
 * add_swap_count_continuation - called when a swap count is duplicated
 * beyond SWAP_MAP_MAX, it allocates a new page and links that to the entry's
 * page of the original vmalloc'ed swap_map, to hold the continuation count
 * (for that entry and for its neighbouring PAGE_SIZE swap entries).  Called
 * again when count is duplicated beyond SWAP_MAP_MAX * SWAP_CONT_MAX, etc.
 *
 * These continuation pages are seldom referenced: the common paths all work
 * on the original swap_map, only referring to a continuation page when the
 * low "digit" of a count is incremented or decremented through SWAP_MAP_MAX.
 *
 * add_swap_count_continuation(, GFP_ATOMIC) can be called while holding
 * page table locks; if it fails, add_swap_count_continuation(, GFP_KERNEL)
 * can be called after dropping locks.
 */
int add_swap_count_continuation(swp_entry_t entry, gfp_t gfp_mask)
{
	struct swap_info_struct *si;
	struct page *head;
	struct page *page;
	struct page *list_page;
	pgoff_t offset;
	unsigned char count;

	/*
	 * When debugging, it's easier to use __GFP_ZERO here; but it's better
	 * for latency not to zero a page while GFP_ATOMIC and holding locks.
	 */
	page = alloc_page(gfp_mask | __GFP_HIGHMEM);

	si = swap_info_get(entry);
	if (!si) {
		/*
		 * An acceptable race has occurred since the failing
		 * __swap_duplicate(): the swap entry has been freed,
		 * perhaps even the whole swap_map cleared for swapoff.
		 */
		goto outer;
	}

	offset = swp_offset(entry);
	count = si->swap_map[offset] & ~SWAP_HAS_CACHE;

	if ((count & ~COUNT_CONTINUED) != SWAP_MAP_MAX) {
		/*
		 * The higher the swap count, the more likely it is that tasks
		 * will race to add swap count continuation: we need to avoid
		 * over-provisioning.
		 */
		goto out;
	}

	if (!page) {
2805
		spin_unlock(&si->lock);
H
Hugh Dickins 已提交
2806 2807 2808 2809 2810
		return -ENOMEM;
	}

	/*
	 * We are fortunate that although vmalloc_to_page uses pte_offset_map,
S
Seth Jennings 已提交
2811 2812
	 * no architecture is using highmem pages for kernel page tables: so it
	 * will not corrupt the GFP_ATOMIC caller's atomic page table kmaps.
H
Hugh Dickins 已提交
2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837
	 */
	head = vmalloc_to_page(si->swap_map + offset);
	offset &= ~PAGE_MASK;

	/*
	 * Page allocation does not initialize the page's lru field,
	 * but it does always reset its private field.
	 */
	if (!page_private(head)) {
		BUG_ON(count & COUNT_CONTINUED);
		INIT_LIST_HEAD(&head->lru);
		set_page_private(head, SWP_CONTINUED);
		si->flags |= SWP_CONTINUED;
	}

	list_for_each_entry(list_page, &head->lru, lru) {
		unsigned char *map;

		/*
		 * If the previous map said no continuation, but we've found
		 * a continuation page, free our allocation and use this one.
		 */
		if (!(count & COUNT_CONTINUED))
			goto out;

2838
		map = kmap_atomic(list_page) + offset;
H
Hugh Dickins 已提交
2839
		count = *map;
2840
		kunmap_atomic(map);
H
Hugh Dickins 已提交
2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852

		/*
		 * If this continuation count now has some space in it,
		 * free our allocation and use this one.
		 */
		if ((count & ~COUNT_CONTINUED) != SWAP_CONT_MAX)
			goto out;
	}

	list_add_tail(&page->lru, &head->lru);
	page = NULL;			/* now it's attached, don't free it */
out:
2853
	spin_unlock(&si->lock);
H
Hugh Dickins 已提交
2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882
outer:
	if (page)
		__free_page(page);
	return 0;
}

/*
 * swap_count_continued - when the original swap_map count is incremented
 * from SWAP_MAP_MAX, check if there is already a continuation page to carry
 * into, carry if so, or else fail until a new continuation page is allocated;
 * when the original swap_map count is decremented from 0 with continuation,
 * borrow from the continuation and report whether it still holds more.
 * Called while __swap_duplicate() or swap_entry_free() holds swap_lock.
 */
static bool swap_count_continued(struct swap_info_struct *si,
				 pgoff_t offset, unsigned char count)
{
	struct page *head;
	struct page *page;
	unsigned char *map;

	head = vmalloc_to_page(si->swap_map + offset);
	if (page_private(head) != SWP_CONTINUED) {
		BUG_ON(count & COUNT_CONTINUED);
		return false;		/* need to add count continuation */
	}

	offset &= ~PAGE_MASK;
	page = list_entry(head->lru.next, struct page, lru);
2883
	map = kmap_atomic(page) + offset;
H
Hugh Dickins 已提交
2884 2885 2886 2887 2888 2889 2890 2891 2892

	if (count == SWAP_MAP_MAX)	/* initial increment from swap_map */
		goto init_map;		/* jump over SWAP_CONT_MAX checks */

	if (count == (SWAP_MAP_MAX | COUNT_CONTINUED)) { /* incrementing */
		/*
		 * Think of how you add 1 to 999
		 */
		while (*map == (SWAP_CONT_MAX | COUNT_CONTINUED)) {
2893
			kunmap_atomic(map);
H
Hugh Dickins 已提交
2894 2895
			page = list_entry(page->lru.next, struct page, lru);
			BUG_ON(page == head);
2896
			map = kmap_atomic(page) + offset;
H
Hugh Dickins 已提交
2897 2898
		}
		if (*map == SWAP_CONT_MAX) {
2899
			kunmap_atomic(map);
H
Hugh Dickins 已提交
2900 2901 2902
			page = list_entry(page->lru.next, struct page, lru);
			if (page == head)
				return false;	/* add count continuation */
2903
			map = kmap_atomic(page) + offset;
H
Hugh Dickins 已提交
2904 2905 2906
init_map:		*map = 0;		/* we didn't zero the page */
		}
		*map += 1;
2907
		kunmap_atomic(map);
H
Hugh Dickins 已提交
2908 2909
		page = list_entry(page->lru.prev, struct page, lru);
		while (page != head) {
2910
			map = kmap_atomic(page) + offset;
H
Hugh Dickins 已提交
2911
			*map = COUNT_CONTINUED;
2912
			kunmap_atomic(map);
H
Hugh Dickins 已提交
2913 2914 2915 2916 2917 2918 2919 2920 2921 2922
			page = list_entry(page->lru.prev, struct page, lru);
		}
		return true;			/* incremented */

	} else {				/* decrementing */
		/*
		 * Think of how you subtract 1 from 1000
		 */
		BUG_ON(count != COUNT_CONTINUED);
		while (*map == COUNT_CONTINUED) {
2923
			kunmap_atomic(map);
H
Hugh Dickins 已提交
2924 2925
			page = list_entry(page->lru.next, struct page, lru);
			BUG_ON(page == head);
2926
			map = kmap_atomic(page) + offset;
H
Hugh Dickins 已提交
2927 2928 2929 2930 2931
		}
		BUG_ON(*map == 0);
		*map -= 1;
		if (*map == 0)
			count = 0;
2932
		kunmap_atomic(map);
H
Hugh Dickins 已提交
2933 2934
		page = list_entry(page->lru.prev, struct page, lru);
		while (page != head) {
2935
			map = kmap_atomic(page) + offset;
H
Hugh Dickins 已提交
2936 2937
			*map = SWAP_CONT_MAX | count;
			count = COUNT_CONTINUED;
2938
			kunmap_atomic(map);
H
Hugh Dickins 已提交
2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956
			page = list_entry(page->lru.prev, struct page, lru);
		}
		return count == COUNT_CONTINUED;
	}
}

/*
 * free_swap_count_continuations - swapoff free all the continuation pages
 * appended to the swap_map, after swap_map is quiesced, before vfree'ing it.
 */
static void free_swap_count_continuations(struct swap_info_struct *si)
{
	pgoff_t offset;

	for (offset = 0; offset < si->max; offset += PAGE_SIZE) {
		struct page *head;
		head = vmalloc_to_page(si->swap_map + offset);
		if (page_private(head)) {
2957 2958 2959 2960
			struct page *page, *next;

			list_for_each_entry_safe(page, next, &head->lru, lru) {
				list_del(&page->lru);
H
Hugh Dickins 已提交
2961 2962 2963 2964 2965
				__free_page(page);
			}
		}
	}
}